Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

Design of Fully Automatic Drone Parachute System with Temperature Compensation Mechanism for Civilian and Military Applications

Design of Fully Automatic Drone Parachute System with Temperature Compensation Mechanism for... Hindawi Journal of Advanced Transportation Volume 2018, Article ID 2964583, 11 pages https://doi.org/10.1155/2018/2964583 Research Article Design of Fully Automatic Drone Parachute System with Temperature Compensation Mechanism for Civilian and Military Applications 1 2 2 Basem AL-Madani, Marius Svirskis, Gintautas Narvydas, 3 3 Rytis Maskeli0nas , and Robertas DamaševiIius Computer Engineering Department, College of Computer Science and Engineering, King Fahd University of Petroleum and Minerals, P.O. Box 1195, Dhahran 31261, Saudi Arabia Department of Automation, Kaunas University of Technology, StudentJ 48, 111, 51367 Kaunas, Lithuania Department of Multimedia Engineering, Kaunas University of Technology, Barsausko 59, A338, Kaunas, Lithuania Correspondence should be addressed to Robertas Damaˇsevici ˇ us; robertas.damasevicius@ktu.lt Received 23 May 2018; Accepted 4 November 2018; Published 15 November 2018 Academic Editor: Yair Wiseman Copyright © 2018 Basem AL-Madani et al. is Th is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. Application of Unmanned Aerial Vehicles (a.k.a. drones) is becoming more popular and their safety is becoming a serious concern. Due to high cost of top-end drones and requirements for secure landing, development of reliable drone recovery systems is a hot topic now. In this paper, we describe the development of a parachute system with fall detection based on accelerometer-gyroscope MPU – 6050 and fall detection algorithm based on the Kalman filter to reduce acceleration errors while drone is yfl ing. We developed the compensation algorithm for temperature-related accelerometer errors. The parachute system tests were performed from a small height on a soft surface. Later, the system was tested under real-world conditions. The system functioned effectively, resulting in parachute activation times of less than 0.5s. We also discuss the civilian and military applications of the developed recovery system in harsh (high temperature) environment. 1. Introduction be assigned to perform. Statistics of accidents during the operation of an Unmanned Aerial Vehicle (UAV) exhibits Unmanned Aerial Vehicles (UAVs), popularly known as a significantly higher accident rate compared to piloted drones, are autonomously or remotely operated aircrasft , aircrasft [16]. This poses severe limitations on the possible which can yfl without on-board human pilot, while being adoption of unmanned systems particularly in the civilian operated from the ground [1]. Currently, drones are for air space. During the flight, the drones may face a variety multiple purposes such as for entertainment (as a toy [2]), of problems, both internal (e.g., an electrical circuit failure, for civilian applications (for example, monitoring soil erosion broken connection, or mechanical damage) and external [3] and wildlife [4], remote sensing [5], surveying and pho- (interference of an external obstacle such as birds or hostile togrammetry [6], smart agriculture [7], disaster surveying force). Due to these problems, drones can become unman- and management [8], forestry [9], energy harvesting [10], ageable, fall down, and land on hard surface. In such cases, and housing renovation [11]), scientific applications [12], and not only the drone and its carrying equipment, but also military applications (surveillance and reconnaissance [13] the objects below (e.g., people or valuable property) can be and attacking the enemy [14, 15]). damaged resulting in large financial losses in addition to lost equipment and information. As the evolution of drone technology continues, its battery capacity and the range of the flight increase, therefore Hence, one important challenge is the implementation also increasing the range and scope of tasks drones can of technologies, which can deal gracefully with failures and 2 Journal of Advanced Transportation ensure safe operation even in the event of engine loss. In such as parachute, network, or hook-trapping. The descent ultralight aviation, emergency parachutes have become an of these systems can be automated using evolutionary algo- important tool to avert damages and casualties caused by rithms that help prepare the drone for landing, braking before aircraft failures. Parachute recovery is particularly suited to landing. tactical fixed wing UAV systems that require a high degree Recognition of a critical failure (due to internal mal- of mobility by allowing air vehicle recovery onto unprepared function or interaction with an external object or force) is terrain [17]. A higher attrition rate for UAVs compared to a distinctive problem in itself [27, 28]. Timely detection of manned aircraft can be due to both physical aircraft and critical failures can be used to enable activation of drone human operator factors. Given that the aircraft does not have recovery system in order to save an aircraft and/or sensitive to accommodate a human being, it can be made smaller. By data from being destroyed or falling into adversary hands. making the aircraft smaller, it is easier to stay undetected in Solutions include hardware redundancy to detect faulty sensors or equipment [29] and analytical redundancy using, the sky for longer time compared to a larger aircra.ft However, due to the size of the UAV, it will encounter dieff rent envi- e.g., fuzzy logic, to detect faulty sensor outputs [30], or ronmental issues that designers and operators may not have capturing an external view of an aircraft and subsequently had experience with in manned aircraft, thus requiring the applying image processing techniques to detect damage to the deployment of emergency recovery system such as parachute. body or parts of an aircraft [28]. However, due to the weight, Such problems could be avoided by using the drone- size, and cost limitations, it is not always possible to replicate mounted parachute system, which would expel parachute or put additional equipment on a small UAV such as a drone. during the fall and ensure safe landing. As in the course In case of critical failure determined, the autonomous of damage or hostile attack such as attempt of seizure, the parachute deployment systems are being the primary drone connection of drone to the control centre may be lost, recovery method now [31–33]. Autonomous Emergency sys- and itmay not bepossibleto activate theparachuteby tems are responsible for deploying the parachute in case of a sending a signal from the control centre. In addition, the critical failure (e.g., lost radio control signal or electrical or drone on-board sensors could detect a possible damage engine failure) [34]. The main advantage of the parachute- or loss of equipment during flight much faster and more based drone recovery method is that the UAV may be reliably, especially at high altitudes. Therefore, an automatic retrieved at any time anywhere (although an accurate point parachute system is a desired solution for the protection of landing is difficult). In addition, if a parachute is used the UAV [18]. It must also be reliable, to eliminate possible as the primary recovery method, it may also be used as mistakes and unintentional firing of the parachute [19], the emergency recovery method saving additional cost and because unnecessary parachuting at low altitude can damage weight. These systems are designed to safely locate aircrasft in objects on the ground. Avoiding such accidents requires emergency situations. Usually, the UAV ifl es autonomously to accurate identification of the free falling state regardless of a preselected site and the parachute is deployed ensuring safe theaxis and angleof thefall [20], but considering varying recovery. Most parachutes are parafoil or cruciform shaped, airflow the drone is subject to during the flight is a consider- stored in the fuselage. In cases where external sensitive able challenge. Analysing drone on-board sensor parameters equipment is under the fuselage, the parachute is deployed (such as acceleration [21]) can help to avoid false parachute from the belly and the vehicle is inverted so that the parts deployment, while it still ensures the protection of the drone under the fuselage are protected from the impact. A more from hitting the ground. complicated system for protecting the vehicle structure or any Operational and exploration requirements for UAVs sensitive parts is the combination of a parachute (deployed impose in particular cases the use of rescue and recovery normally on top of the fuselage) and an airbag which is systems. These systems could possess a large range of con- inflated under the fuselage a few seconds before touchdown. structive solutions, such as recover parachutes [22], airbags The airbag may alsobeused tokeep the UAV aofl at when [23], safety nets [24], and pneumatic cushions boats [25]. landed on water. The main problem in parachute recovery The rescue and recovery systems must allow a safe landing is the transition from horizontal to vertical motion of the in case of emergencies and could become a solution to a vehicle, which starts as soon as the parachute is deployed. standardization regarding landing in extreme weather and Then, although parachute steering is possible, the oscillatory geographical conditions [25]. movements of the vehicle present problems in recovery such The aim of this paper is to develop and experimentally as inaccurate point landing or touching down at different validate a drone-mounted parachute system capable of timely angles which may damage the vehicle structure [35]. parachuting and safely landing the drone in case of an Alizadeh et al. [23] suggested using airbag systems as accident, attack, or loss of control. a shock absorber for UAVs to assist with rapid parachute landings. An airbag system is activated on touchdown to absorb the kinetic energy during the impact with the ground. 2. State-of-the-Art Bleier et al. [34] presented the risk assessment models for generation of flight paths, which support the automatic Recovery of self-propelled yin fl g drones in a limited space is a rather difficult task [26]. During the halt, problems such deployment of emergency parachutes for Unmanned Aerial as hovering before landing, precise tracking of the UAV, Vehicles in emergencies due to loss of propulsion. Based on a risk analysis of the area underneath the flight path, and changing weather conditions are encountered. In this situation, there are several possible options for dropdowns suitable deployment positions are identified, which minimize Journal of Advanced Transportation 3 thechanceofendangering humans on theground, property 3. Method damage, and loss of the air vehicle. Du [36] introduced 3.1. Parachute Design Considerations. The purpose of a simulation model of a certain kind UAVs for UAV recovery parachute is to decrease speed while maintaining stability of system design. a falling vehicle or payload. The aerodynamic and stability Guo et al. [37] provided six degrees of freedom (DOF) characteristics of the parachute are determined by its geomet- flight dynamics model of UAV and parachute recovery ric features, which influence opening force and movement dynamic model for a quantitative prediction of the motion trajectory [43]. UAVs are subject to increasing demands and process of UAV and recovery system and predicted dangerous are committed to achieving maximum operational reliabil- situations at recovery stage. Kim et al. [31] predicted the ity. Parachutesystems are used to ensurethereliability of parachute deployment for landing at the desired point using operation and the safety of the machine. It is important to the neural network and the flight conditions such as the know the characteristics of the parachute for proper landing. deployment position, UAV’s velocity, and wind velocity as Parachutes can be of different types, and their parameters of input data sets, and landing points such as the cross range resistance to wind also vary greatly. Inflatable parachute is and the down range position as output data sets. Li et al. [32] suitable for slowing down at high altitude, where there is a analysed the motion characteristics of the parachute-UAV very rare atmosphere. Parachutes of this type are intended for combination in the whole recycling process and a comparison lowering scientific equipment and measure the direction of a of the calculated results to the experimental results to verify horizontal moving wind. Therefore, the release of a parachute the reliability of the numerical simulation. Shao et al. [22] must be fast and efficient. The effectiveness of the parachute evaluated a model of the UAV-parachute system, with wind spread depends on its size. A simple parachute, which does fields and a control strategy for recovery. not need to be controlled gradually, is effective only up to Shyu andHsiao [33] describedthe development of 1.5 m in diameter. Parachutes of this type are expanding mini-UAV that has the capabilities of bungee launch and very quickly; they can be used for discharging the opening parachute recovery. Yong and Li [38] studied the optimal container, from which the parachute simply falls out. Also, a parachute decelerating trajectory of mini-UAVs, which is spring exhaust mechanism can be used. solved by sequential quadratic programming method and Larger parachutes must withstand much higher overload genetic algorithm, while the results are used to guide due to high air resistance during take-off. In order to avoid research of parachute design. Zeng and Cai [39] introduced breaking of the cord, due to the occurrence of a shock, it is the parachute recovery system composed of fuselage, door, necessary to tighten the ropes before the parachute is pulled hanger, segregator, shooting system, main parachute, drogue out. One of the options is to use a rocket engine that pulls parachute, unlocking system of doors, and binding system. the parachute forward. However, such a system takes up quite The parachute system was optimized by wind tunnel tests a lot of space, which is difficult and not practical in small for the UAV being in unfavoured states by air-drop experi- aircra.ft Another option is to use a small parachute that drags ment. out a larger parachute. All variations must have the appro- Morgan et al. [40] explore the build and design of a priate fall speed during the launch of the parachute. When ballistic parachute recovery system. This system will monitor designing a parachute, several basic criteria need to be taken several variables in real-time to determine whether or not into consideration: simplicity of design and construction, the aircraft is operating in a safe environment. The elements resistance to tensile stress, and high wind resistance. observed are main battery voltage, current GPS coordinates, and current acceleration. If the system determines that the 3.2. Mathematical Model of Parachute Fall. It is difficult to aircraft is operating in an unsafe environment, the recovery model the dynamics of parachute ejection, opening, and system will cut main power and deploy a ballistic parachute falling. Usually some simplifications are adopted; for exam- to guide the aircraft safely to the ground. ple, the parachute is considered to be a rigid body which is Cristian and Codrea [41] designed parachute recovery affected by aerodynamic drag, strains the connection riser, and landing attenuation system for a military reconnaissance and transmits a force to the payload with six degrees of drone. The recovery system is able to recover the air vehicle freedom (see Figure 1). after the complete mission when the vehicle has landed in The system of equations presented here is based on the rough terrain at altitudes from sea level to 3000 m; recover previous worksof[43–46] andis denfi ed asfollows: the drone during uncontrolled flight conditions; and serve as safety device to prevent the air vehicle leaving the boundaries 𝑓 =𝑀 of the zone. Safe landing problems are also relevant in rocket systems 𝑓 =𝑋−𝑔𝑚 sin𝜃 [42]. Hybrid rockets can be used to control inflatable wings. (1) The advantage of such a system against a simple parachute 𝑓 =𝑍+𝑔𝑚 cos𝜃 system is that the landing is controlled and the inflatable 𝑓 =𝜃−𝛾−𝛼 wings help totravel tothe desired landing location. The inflatable wings occupy a much smaller area than the folds here, of similar dimensions. However, this type of wing system is difficult to reconcile, the wings are not very stable, they need 𝑀= 𝜌𝑉 𝑆 𝐶 +𝐹 cos𝛼+𝐹 sin𝛼 𝑝 𝑀 𝑅 𝑅 an inflated compressed gas control equipment. 𝑝 2 4 Journal of Advanced Transportation The angle of attack is given by 󵄨 󵄨 󵄨 󵄨 −1 󵄨 𝑧 󵄨 󵄨 󵄨 𝛼= tan (󵄨 󵄨 ) (7) 󵄨 󵄨 󵄨 󵄨 󵄨 󵄨 The angle of sideslip is derived by calculating - Tension Force 󵄨 󵄨 󵄨 󵄨 󵄨 𝑦 󵄨 −1 󵄨 󵄨 󵄨 󵄨 𝛽= sin ( ) (8) - Damping 󵄨 󵄨 󵄨 󵄨 󵄨 𝑉 󵄨 󵄨 󵄨 setting 3.3. Calculation of Drone Descent Speed. Drone speed is calculated in several steps. In order to know what acceleration the object moves, regardless of the accelerometer position, the sum of the acceleration values for all axes is expressed as 2 2 2 𝐺 =√𝐺 +𝐺 +𝐺 (9) 𝑥 𝑦 𝑧 where 𝐺 , 𝐺 , 𝐺 are the accelerometer values of the 𝑥 𝑦 𝑧 accelerometer axes (x, y, z). Assuming that v = 0 m / s, we can calculate the total fall speed as follows: Figure 1: Model of parachute system. 𝑡 𝑘𝑟 𝑘𝑟 󵄨 󵄨 →󳨀 󵄨 󵄨 󵄨 󵄨 (10) 𝑉 =∫ 󵄨 𝐺 󵄨 = Δ𝑡 ∑𝐺 𝑘𝑟 𝑧 󵄨 󵄨 2 󵄨 󵄨 𝑋= 𝜌𝑉 𝑆 𝐶 −𝐹 cos𝛼 𝑝 𝑝 𝑥 𝑅 where𝑡 is the time after which it is checked or the fall speed 𝑘𝑟 𝑍= 𝜌𝑉 𝑆 𝐶 −𝐹 sin𝛼 has not exceeded the set limit. 𝑝 𝑝 𝑧 𝑅 To eject the parachute emission, we apply the condition (2) that it must reach a certain speed (1.5 m/s) or more within a specified time interval (0.25 s). Otherwise, the parachute where M is the moment, X, Y,and Z are the body forces, 𝑉 will not be ejected. Under such conditions, the drone is not is velocity, S is parachute surface area, 𝐹 is the riser force allowed to reach a dangerous fall speed. (tension),𝛼 is attack angle, and𝛽 is sideslip angle. The deceleration of the parachute is calculated as follows: 3.4. Parameter Control. In order to understand the causes of 𝐹 −𝐹 −𝑚 𝑔 sin𝛾 𝑑𝑉 𝑅 𝐷 𝑝 a drone accident and to identify the distinctive features, it is (3) necessary to analyse previous accidents aiming to find possi- (𝑚 +𝑚 ) 𝑝 𝑑 ble solutions to problems. Having evaluated the real situations where 𝑚 is the parachute mass, 𝑚 is the drone mass, 𝐹 and in order for the algorithm to be able to distinguish the 𝑝 𝑑 𝑅 is riser force, 𝐹 is parachute drag force, and 𝛾 is flight path fall data from the movement data, the fall parameters must angle, and a is acceleration. be properly described. By measuring accelerometer readings The drag force of the parachute is calculated as follows: and evaluating parameters with a certain parameter limit, it is possible to detect the moment of falling. The system must clearly detect the drop and the rise 𝐹 = 𝑉 𝐷𝑆 (4) 𝐷 𝑝 regardless of the drone orientation in the space [47]. When knowing the exact moment from when the object began to where𝐷 is drag coefficient, 𝑆 is projected parachute area,𝑉 𝑝 𝑝 fall and when it reached the ground, it is possible to calculate is parachute velocity, 𝜌 is air density, and 𝜂 is an efficiency the height of the fall. Detection is also relevant to protecting factor. electronics from potential damage during impact. Detecting As a result, the following system of equations is con- a fall quickly is especially important in cases where the structed. operation has a very low height. This requires an optimized The falling speed is from the body velocities in three axes: program code and an appropriate microcontroller capable of 2 2 2 processing information as quickly as possible. 𝑉=√ 𝑉 +𝑉 +𝑉 (5) 𝑥 𝑦 𝑧 To determine the position and height change it is nec- essary to process the received sensor readings and to reject The deceleration is calculated as the derivative of the velocity incorrect values and calculate speeds at certain time intervals. vector: Position tracking algorithms with accelerometers can be 𝑉 𝑑𝑉 /𝑑𝑡+𝑉 𝑑𝑉 /𝑑𝑡+𝑉 𝑑𝑉 /𝑑𝑡 𝑑𝑉 𝑥 𝑥 𝑦 𝑦 𝑧 𝑧 used in situations where high precision is not required. The (6) 𝑉 integration algorithm is ideal for low-cost devices due to its Distance to impact Fall velocity & angle 𝑑𝑡 𝜂𝜌 𝑑𝑡 𝑥𝑦 𝑑𝑡 𝑥𝑦 Journal of Advanced Transportation 5 1.2 1.2 1.15 1.15 1.1 1.1 1.05 1.05 0.95 0.95 0.9 0.9 0.85 0.85 0.8 0.8 0 10 20304050 010 20 30 40 50 Temperature ( C) Temperature ( C) 2 ∘ Figure 2: Dependence of the accelerometer Z-axis values (1g – 9.81 m/s ) upon the temperature during rapid (5 C/min) heating (left) and cooling (right). simplicity. Also, for all speeds, all calculations are performed performed 10 times for each axis of the sensor, in a position using only oa fl ting point operations. The data is processed that maximizes the values of that axis. using a low frequency filter. Several different sensors can be A collection of values has been created in the processing used to increase the positioning accuracy. environment, which reads the values through a serial data output and writes them to a text file. The program reads 3.5. Calibration of Accelerometer. Accurate accelerometer sensor values and checks the conditions to avoid false values, values are particularly important for drone speed detection. checking whether the value arrives and whether it has more Due to temperature, accelerometer chip manufacturing inac- than 30 “char” elements. If conditions are met, values are curacies, and dieff rent stresses occurring during soldering written to the text document. This cycle is repeated until the and assembly, deviation of accelerometer values occurs. Mini- program is stopped (Figure 2). mal and maximum values of the scanned values are processed Figure 2 shows the deviations of the accelerometer val- by the Kalman lfi ter [48] (the delay in the submission of ues depending upon the temperature values. We can see data is not affected during the calibration). The filtered values that the acceleration values vary greatly. We assumed the are compared and recorded at the highest. Subsequently, the accelerometer can not warm up or cool down evenly due to subroutine is stopped, the accelerometer is overturned, the rapid heating or cooling (5 C/min). Therefore, it is necessary accelerometer axis being measured in the opposite direction, to significantly reduce the temperature variation in the and then the scan is continued. The same calibration steps are temperature control chamber. By reducing the temperature repeated for each axis. change to 1 C per minute, we obtained the following heating Accelerometers and gyros have high noise and temper- and cooling charts (Figure 3). ature variations. Therefore, it is not enough to take the From the graph (Figure 3) using linear regression, we values of the sensor data alone, but we also need to know have derived the following linear functions of the change the environmental conditions do the sensor values could be of acceleration during the cooling and heating process as compensated for temperature compensation. It is necessary follows: to measure the sensor values at different temperatures in 𝐺 = −0.0047𝑇+1.1031 order to determine possible errors. (11) A device for adjusting the temperature was made for 𝐺 = −0.0056𝑇+1.0967 calibration of the accelerometer. The thermoelectric heating and cooling element was used for this purpose, operating where𝐺 is the acceleration measured by the accelerometer’s under the Peltier effect principle. Also used are the temper- Z-axis. ature sensor DS18B20, the radiator for excessive heat or cold The initial and final values of the graphs are similar. of the thermoelectric element, a fan, a 12V power supply, Therefore, we can assume that during the cooling time the a bidirectional motor control TB6612FNG, and an Arduino accelerometer values need to be further elaborated. Since, microcontroller. by heating, the values change evenly without any significant By inserting the accelerometer into the temperature distortion, we assume that deviations of the Z-axis of the control chamber, rfi st of all, the chamber is positioned so sensor from the norm can be compensated by using this that the axis measured by the sensor is perpendicular to the function. The same test is performed with other axes (see the ground. The fan is mounted on the stand so that it does not results presented in Figures 4 and 5). touch the cooling system at all and does not aec ff t the sensor’s From Figures 4 and 5, we can note that the temperature- measurement data. Then, the sensor is cooled to 0 C, then induced variations on the X-axis and the Y-axis are con- ∘ ∘ heated to 50 C, and then cooled again to 0 C. The process is siderably smaller than the ones on the Z-axis. The Z-axis Acceleration (g) Acceleration (g) 6 Journal of Advanced Transportation 1.2 1.2 1.15 1.15 1.1 1.1 1.05 1.05 1 1 0.95 0.95 0.9 0.9 0.85 0.85 0.8 0.8 0 10 20304050 010 20 30 40 50 2 ∘ Figure 3: Dependence of the accelerometer Z-axis values (1g – 9.81 m/s ) upon the temperature during slow (1 C/min) heating (left) and cooling (right). 1.2 1.2 1.15 1.15 1.1 1.1 1.05 1.05 0.95 0.95 0.9 0.9 0.85 0.85 0.8 0.8 0 10 20304050 010 20 30 40 50 2 ∘ Figure 4: Dependence of the accelerometer X-axis values (1g – 9.81 m/s ) upon the temperature during slow (1 C/min) heating (left) and cooling (right). temperature compensation is derived from the heating value protocol between the receiver and the master controller. function, and the temperature compensation formula is as The main controller has several sensors: an accelerometer, a follows: magnetometer, a barometer, and a gyroscope. In order for the drone to yfl stably, it was necessary for this controller to 𝐺 =𝐺 +0.0047𝑇−0.1031 (12) 𝑧 𝑝𝑐𝑜𝑚 𝑧 adjust the PID parameters. The settings have been adjusted manually. where 𝐺 is the compensated acceleration value, 𝐺 is 𝑧 𝑝𝑐𝑜𝑚 𝑧 the measured value of the accelerometer Z-axis, and T is the 4.2. Electronics. A system for unlocking the parachute ejec- temperature. tion mechanism was created (Figure 8), which deploys the From the graphs of compensated values (Figure 6), we can spring holding rope. This system is particularly lightweight conclude that the deviations of the values are compensated. and has a small additional space, since it no longer requires However, due to differences in heating and cooling (at the a heavy servo motor. The Nichrome wire is used to burn same temperature), the values are compensated to a certain the cord. A MOS type transistor is used to prevent the cord limit, until the maximum deviation reaches 5%. The deviation from slipping off when the current is started. Transistor of the values before compensation was 14%. bandwidth is controlled by a PWM type signal. The system is managed by the Arduino microcontroller, which processes 4. Drone Design and Testing the sensor data, and controls the outputs. A battery of 110 4.1. Settings. Drone testing was performed under real condi- mAh lithium polymer has been selected for power supply. tions (see Figure 7). The drone frame consisted of a balsam Its charging system has an integrated battery charger with tree reinforced with a glass fibre. This type of frame is battery protection from excessive discharge. The battery is charged through the micro USB connector. The RGB LED extremely lightweight and has low production and repair costs. The electric motor bearing structures consisted of is used for indicating the operation. The LED at the start of aluminium hollow bars. The Emax company controller was the system, in colour, indicates the level of the battery voltage selected for drone engine management. The drone is con- and informs about the operation of the system and the fall trolled using the 9-channel receiver using the S-Bus type detection. Journal of Advanced Transportation 7 1.2 1.2 1.15 1.15 1.1 1.1 1.05 1.05 0.95 0.95 0.9 0.9 0.85 0.85 0.8 0.8 010 20 30 40 50 0 10 20304050 2 ∘ Figure 5: Dependence of the accelerometer Y-axis values (1g – 9.81 m/s ) upon the temperature during slow (1 C/min) heating (left) and cooling (right). 1.2 1.2 1.15 1.15 1.1 1.1 1.05 1.05 1 1 0.95 0.95 0.9 0.9 0.85 0.85 0 10 20304050 0 10 20304050 2 ∘ Figure 6: Dependence of the compensated accelerometer Z-axis values (1g – 9.81 m/s ) upon the temperature during slow (1 C/min) heating (left) and cooling (right). Figure 8: Parachute system electronic circuit diagram: 1 – charging unit; 2 - switch; 3 – Lithium polymer battery; 4 - microcontroller; 5 – accelerometer - gyroscope; 6 – LED diode; 7 – MOS type transistor; Figure 7: Ready-to-yfl drone. 8 – nichrome wire. 4.3. Control Algorithm. For fall detection, we have adopted an approach that is based on the detection of the falling state axis values are lfi tered using a Kalman filter. Using the filtered of the drone. The approach in itself in not new [49] and we data, the combined acceleration of all axles is calculated. selected it due to its simplicity, however, we have improved Then, the condition is checked whether the speed limit is the approach by introducing compensation for temperature xfi ed. If the limit is xfi ed - the time from the moment of caused variability. The parachute system program (Figure 9) capture is measured. After a limited time, the fall acceleration measures the battery voltage, then the accelerometer and is checked again. If the condition ts, fi the parachute is ejected. gyroscope data. The Z-axis data on the accelerometer is Also, one of the main conditions for detecting the fall state or compensated by the temperature-dependent deviation. The the unstable operation of a drone is the measurement of the 8 Journal of Advanced Transportation START Start calculating time, set first variable to 1 No Yes Is voltage Initiate sensor >3.8V? readout Read accel. data Calculate acceleration on all axis Is Yes No Initiate angle accel. Value Is this an initial equal to fall accel. Readout? readout value? No Yes Read angle data No Yes Is angle >90 Drop the parachute Yes No End of First variable is Deadlock on? programme marked as 0 Figure 9: Parachute system operation algorithm. angle of inclination. When the limit value (at 90 ) is exceeded, (the system was thrown onto a soft surface). First of all, the parachute is ejected. If no condition is satisfied, the cycle the simple fall detection algorithm was tested: when the is repeated until the condition is satisfied or the shutdown average of the total values of the acceleration was 15 and button is depressed. the threshold was applied to it. When crossing the limit, the parachute system was triggered. The system worked 4.4. Parachute Exhaust System. The parachute was made reliably. However, when placed on the drone, no fall was from a rugged nylon material. Parachute dimensions are recorded due to excessive vibrations. Using the Kalman filter for this algorithm, the system worked efficiently. The fall chosen such that a 2 kg drone does not fall faster than 5 m/s. Under the selected dimensions, a 16-segment dome-shaped was recorded in 0.5 s. In order to prevent the unintentional parachute with overlock stitches was made (Figure 10). The ejection, the fall acceleration was checked for a second time diameter of the parachute is 1.45 m. aer ft a specified period of time. Only in accordance with The parachute exhaust housings (Figure 11) were made the second measurement condition the parachute is to be of PVC pipe, glass br fi e board, and a bottom-mounted opened. During critical conditions, when vibrations have mounting piece made of three-dimensional printing. extremely high amplitude and frequency, both Kalman and the compliant filter failed to properly lfi ter the errors. As a result, the fall was sometimes erroneously detected. However, 4.5. Testing. To determine the efficiency of the parachute system, we attempted to detect a fall without the drone itself in real conditions, we have not detected any false falls. The Journal of Advanced Transportation 9 Figure 12: System testing under real conditions. Figure 10: eTh 16-segment, dome-shaped parachute. of different environments, while ensuring the avoidance of obstacles [50]. For example, the meteorology drones collect data on air temperature, humidity, pressure, wind force, radi- ation, etc. and should beabletowithstand theadverse forces of nature such as strong wind gusts. If yfl ing at high altitudes also the temperature factor should be taken into account [51]. There are significant challenges related to predicting tem- perature variations when using drones in Arctic [52], where temperature may fall below -50 C, or in deserts [53], where Figure 11: Parachute system body. temperature may exceed +50 C, but also for more common civilian applications executed during nigh time [54], or for forest fire monitoring [55], when temperatures may fall below operating parachute system during opening stage is depicted or well exceed the range of temperatures most consumer ∘ ∘ in Figure 12. drones are designed for, which is usually 0 C–40 C. In During the tests, an attempt was made to accelerate the this paper we have addressed the challenge of designing a parachute opening, and the parachute was covered with Talc reliable parachute-based recovery system for a drone, which powder from the bonding of the material. A special parachute uses the temperature-compensated accelerometer data for folding technique is used when the parachute is folded in parachute ejection. Note that still many other temperature- such a way as to spill as soon as possible. As a result, folded related challenges for drone designers exist, most notably, parachute can open much faster. However, the minimum the impact of temperature on the battery performance and height required for the parachute to land was about 20 m from characteristics. the ground. 6. Conclusion 5. Evaluation We have proposed a parachute-based system for drone recov- We evaluate the developed UAV parachute system according ery with the accelerator sensor temperature compensation to the requirements for UAV recovery systems formulated by mechanism. The parachute system has been successfully Wyllie [17] as follows: tested in real-life conditions. The fall is detected almost immediately (within 0.5 s). However, it takes much more time (i) Safety: a mandatory requirement during landing. for the parachute to unfold, so at less than 20 m height, the (ii) Protection: protect drone and sensitive on board parachute still may not be able to unfold in time. The system equipment from damage during landing. also includes a mechanism for prevention of unintended parachute deployment. In calculating the rate of fall, the (iii) Accuracy: ensure high accuracy when landing to a resulting acceleration errors, even aeft r compensation, were planned landing point to minimize landing damage. high. The calculated speed data could not be relied upon even (iv) Automation: reduce the number of manual opera- at small intervals, assuming that an initial speed of falling is tions by the drone operator. equal to 0 m/s. The acceleration deviations can be offset by (v) Reliability: ensure predictability of operation. applying the Kalman lfi ter. The accelerometer values depend significantly upon the (vi) Repeatability: ensure that recovery may be repeated environment temperature, so the compensation mechanism many times during lifetime of the drone. based on linear regression was introduced to allow stable Since the areas where drones are used are expanding with new detection of drone fall state during harsh environment con- applications proposed every year, it is important that drones ditions, such as in a desert environment during a hot sum- continue to be used both safely and efficiently in multitude mer. 10 Journal of Advanced Transportation Data Availability systems (UASs), part 2: scientific and commercial applications,” Journal of Unmanned Vehicle Systems, vol.02,no.03,pp. 86–102, The data used to support the findings of this study are available from the corresponding author upon request. [13] C. Paucar,L.Morales, K.Pinto et al., “Use of drones forsurveil- lance and reconnaissance of military areas,” in Developments and Advances in Defense and Security. MICRADS 2018. Smart Conflicts of Interest Innovation, Systems and Technologies, A. Rocha and T. Guarda, Eds., vol.94, pp.119–132,Springer, Cham,2018. The authors declare that they have no conflicts of interest. [14] J. Drew, “Drone strike: Long-range attack UAVs being devel- oped from aerial targets,” Aviation Week & Space Technology, Acknowledgments vol.179, no.4,pp.42-43,2017. The authors acknowledge the support by King Fahd Univer- [15] D. Hambling, “Military tackles the problem and the potential of drones,” Aviation Week & Space Technology, vol. 180, no. 8, pp. sity of Petroleum & Minerals (KFUPM) in publishing this 44–47, 2018. work. [16] G. Wild, J. Murray, and G. Baxter, “Exploring Civil Drone Acci- dents and Incidents to Help Prevent Potential Air Disasters,” References Aerospace, vol. 3, no. 3, p. 22, 2016. [1] E. N. Barmpounakis, E. I. Vlahogianni, and J. C. Golias, “Un- [17] T. Wyllie, “Parachute recovery for UAV systems,” Aircraft Engineering and Aerospace Technology, vol.73, no.6,pp.542– manned Aerial Aircraft Systems for transportation engineering: Current practice and future challenges,” International Journal of 551, 2001. Transportation Science and Technology, vol. 5,no. 3, pp.111–122, [18] K. Graham and S. Cartwright, Feasibility of Parachute Recovery Systems for Small UAVs, University of New South Wales, New South Wales, 2008. [2] C.-H. Poh and C.-K. Poh, “Radio Controlled “3D Aerobatic Air- planes” as Basis for Fixed-Wing UAVs with VTOL Capability,” [19] D. Lim,C. Park, N. H.Kim,S.-H. Kim,and Y. S.Yu, “Fall- Detection Algorithm Using 3-Axis Acceleration: Combination Open Journal of Applied Sciences,vol.04,no. 12,pp.515–521, 2014. with Simple Threshold and Hidden Markov Model,” Journal of Applied Mathematics,vol.2014,ArticleID896030,8pages,2014. [3] S. D’Oleire-Oltmanns, I. Marzolff, K. D. Peter, and J. B. Ries, [20] J. Luo, B. Zhong, and D. Lv, “Fall Monitoring Device for Old “Unmanned aerial vehicle (UAV) for monitoring soil erosion People based on Tri-Axial Accelerometer,” International Journal in Morocco,” Remote Sensing, vol. 4, no. 11, pp. 3390–3416, 2012. of Advanced Computer Science and Applications, vol.6,2015. [4]M.E.Goebel, W. L. Perryman, J.T. Hinke et al., “A small [21] K. Seifert, “Oscar Camacho. Implementing Positioning Algo- unmanned aerial system for estimating abundance and size of rithms Using Accelerometers,” Application Note AN3397, Rev 0, Antarctic predators,” Polar Biology,vol.38,no.5,pp.619–630, Freescale Semiconductor,2007. [22] P. Shao, C.Wu, and S.Ma, “Research onkey problems in [5] K. Whitehead and C. H. Hugenholtz, “Remote sensing of the assigned-point recovery of UAV using parachute,” in Proceed- environment with small unmanned aircraft systems (UASs), ings of the 2013 IEEE International Conference of IEEE Region 10 part 1: a review of progress and challenges,” Journal of Un- (TENCON 2013),pp. 1–4, 2013. manned Vehicle Systems,vol. 02, p. 69,2014. [23] M. Alizadeh, A. Sedaghat, and E. Kargar, “Shape and orifice [6] F.Cuesta, F. Lopez-Rodriguez, and A.Esteban,“A New Blondin optimization of airbag systems for UAV parachute landing,” System for Surveying and Photogrammetry,” Sensors,vol.13,no. International Journal of Aeronautical and Space Sciences,vol. 15, 12, pp. 16894–16914, 2013. no. 3, pp. 335–343, 2014. [7] E. Honkavaara, H. Saari, J. Kaivosoja et al., “Processing and [24] S. Ramasamy, R. Sabatini, and A. Gardi, “A Unified Analytical assessment of spectrometric, stereoscopic imagery collected Framework for Aircraft Separation Assurance and UAS Sense- using a lightweight UAV spectral camera for precision agricul- and-Avoid,” Journal of Intelligent & Robotic Systems,vol.91,no. ture,” Remote Sensing, vol.5,no. 10,pp.5006–5039,2013. 3-4, pp. 735–754, 2018. [8] Z.Xu, J. Yang, C.Peng etal.,“Development ofanUAS for [25] V. Prisacariu, “Recovery System Of The Multi-Helicopter Uav,” post-earthquake disaster surveying and its application in Ms7.0 Review of the Air Force Academy, vol.1, no. 31,pp.91–98, 2016. Lushan Earthquake, Sichuan, China,” Computers & Geosciences, [26] S. Khantsis, Control system design using evolutionary algorithms vol. 68, p. 22, 2014. for autonomous shipboard recovery of unmanned aerial vehicles. [9] L. Tang and G. Shao, “Drone remote sensing for forestry Doctor of Philosophy (PhD), Aerospace, Mechanical and Manu- research and practices,” Journal of Forestry Research,vol. 26, no. facturing Engineering, RMIT University, 2006. 4, pp. 791–797, 2015. [27] C. H.Lo, E. H.Fung, and Y.K.Wong, “Intelligent automatic [10] Z. Zhao, X. Pu, C. Du et al., “Freestanding Flag-Type Triboelec- fault detection for actuator failures in aircra,ft” IEEE Transac- tric Nanogenerator for Harvesting High-Altitude Wind Energy tions on Industrial Informatics, vol.5,no. 1,pp. 50–55, 2009. from Arbitrary Directions,” ACS Nano,vol.10, no.2,pp. 1780– [28] Y. Wiseman, “Device for detection of fuselage defective parts,” 1787, 2016. International Information Institute (Tokyo). Information,vol. 17, [11] A. A. Mustaffa, M. F. Hasmori, A. S. Sarif, N. F. Ahmad, and N. Y. no. 9, pp. 4189–4194, 2014. Zainun, “The Use of UAV in Housing Renovation Identification: [29] D. Rehage, U. B. Carl, and A. Vahl, “Redundancy manage- A Case Study at Taman Manis 2,” IOP Conference Series: Earth ment of fault tolerant aircraft system architectures - Reliability and Environmental Science,vol.140,p.012003,2018. synthesis and analysis of degraded system states,” Journal of [12] K. Whitehead, C. H. Hugenholtz, S. Myshak et al., “Remote Aerospace Science and Technology, vol. 9, no. 4, pp. 337–347, sensing of the environment with small unmanned aircraft 2005. Journal of Advanced Transportation 11 [30] W. A. Kwong, K. M. Passino, E. G. Laukonen, and S. Yurkovich, [48] E. A. Wan and R. Van Der Merwe, “eTh unscented Kalman “Expert supervision of fuzzy learning systems for fault tolerance filter for nonlinear estimation,” in Proceedings of the IEEE Adap- aircraft control,” Proceedings of the IEEE, vol. 83, no.3,pp. 466– tive Systems for Signal Processing, Communications, and Con- 482. trol Symposium (AS-SPCC 2000), pp. 153–158, IEEE, Alberta, Canada, October 2000. [31] I. Kim, S. Park,W.Park,and C.-K.Ryoo, “Prediction of the parachute deploy for landing at the desired point,” in Proceed- [49] K. L. Crandall and M. A. Minor, “UAV fall detection from a dynamic perch using Instantaneous Centers of Rotation and ings of the 2012 Proceedings of SICE Annual Conference, SICE 2012, pp. 1629–1634, IEEE, Akita, Japan, August 2012. inertial sensing,” in Proceedings of the 2015 IEEE International Conference on Robotics and Automation, ICRA 2015,pp.4675– [32] M. Li, H. Zhang, and J. Chen, “Dynamic analysis for parachute 679,USA,2015. recovery system of unmanned aerial vehicle,” Journal of System [50] A. Ivanovas, A. Ostreika, R. Maskeliu¯nas, R. Damaˇsevici ˇ us, Simulation,vol.24, no.6,pp.1303–1307,2012. D. Połap, and M. Wozniak, “Block Matching Based Obstacle [33] L.-S. Shyu and Y.-C. Hsiao, “Mini UAV design and manufacture Avoidance for Unmanned Aerial Vehicle,” in Artificial Intelli- with bungee launched / Parachute recovery,” Applied Mechanics gence and Soft Computing , pp. 58–69, Springer International and Materials,vol.610, pp.97–100, 2014. Publishing, 2018. [34] M. Bleier, F. Settele, M. Krauss, A. Knoll, and K. Schilling, “Risk [51] E. H.Teets Jr., C.J.Donohue,K. Underwood,and J.E. Bauer, assessment of flight paths for automatic emergency parachute “Atmospheric considerations for uninhabited aerial vehicle deployment in UAVs,” IFAC-PapersOnLine, vol.28, no.9,pp. (UAV) flight test planning,” NASA/TM-1998-206541, 1998. 180–185, 2015. [52] M. Ader and D. Axelsson, Drones in Arctic Environments,KTH, [35] W. J. Crowther and K. Prassas, “Post Stall Landing for Field Stockholm, Sweden, 2017. Retrieval of UAVs,” in Proceedings of the 14th Bristol Interna- [53] D. Gallacher, “Applications of micro-UAVs (drones) for desert tional unmanned air vehicle systems conference, 1999. monitoring: current capabilities and requirements,” in Proceed- [36] K. M. Du, “Simulation analysis on the problem of opening main ings of the Kubuqi International Desert Forum 2015,pp. 249–259, parachute used in UAV recovery system,” Applied Mechanics and Materials,vol.390,pp. 57–61,2013. [54] S. R. Herwitz, K. Allmendinger, R. Slye et al., “Nighttime UAV [37] L. Guo, H. Zhang, and M. Tong, “Dynamics analysis on Vineyard Mission: Challenges of See-and-Avoid in the NAS,” parachute recovery of unmanned aerial vehicle,” Journal of in Proceedings of the AIAA 3rd “Unmanned Unlimited” Techni- Nanjing University of Aeronautics and Astronautics,vol. 44,no. cal Conference, Workshop and Exhibit, Chicago, Illinois, USA, 1, pp. 14–19, 2012. September 2004. [38] Z. Yong and L. Li, “Parachute Decelerate Trajectory Optimiza- [55] P. Skorput, S. Mandzuka, and H. Vojvodic, “The use of Un- tion Design Based on Genetic Algorithm,” in Proceedings of the manned Aerial Vehicles for forest fire monitoring,” in Proceed- 2009 International Workshop on Intelligent Systems and Appli- ings of the 58th International Symposium ELMAR, ELMAR 2016, cations, ISA 2009, IEEE, Wuhan, China, 2009. pp. 93–96, Croatia, September 2016. [39] L. Zeng and J. Q. Cai, “Scheme design of a parachute recovery system for UAV,” in Proceedings of the 64th International Astro- nautical Congress 2013, IAC 2013, pp. 8404–8412, China, 2013. [40] A. Morgan and R. Chapman, Ballistic Parachute Recovery Sys- tem for Unmanned Aerial Vehicles (UAVs), Make Magazine, [41] A.-I. Cristian and S. Codrea, “Parachute deployment systems for drones, applied on uav-sact project,” in 15th International Conference eTh Knowledge-Based Organization: Applied Techni- cal Sciences and Advanced Military Technologies,vol.6,pp.260– 266, 2009. [42] C. Sudduch, Design of a Hybrid Rocket/Ina fl table Wing UAV , Oklahoma State University, USA, 2012. [43] G. Guglieri, “Parachute-Payload System Flight Dynamics and Trajectory Simulation,” International Journal of Aerospace Engi- neering, vol. 2012, Article ID 182907, 17 pages, 2012. [44] K.-F. Doherr and H. Schilling, “Nine-degree-of-freedom simu- lation of rotating parachute systems,” Journal of Aircraft ,vol.29, no.5,pp.774–781,1992. [45] V. N.Dobrokhodov,O.A. Yakimenko, and C.J.Junge,“Six- degree-of-freedom model of a controlled circular parachute,” Journal of Aircraft ,vol. 40, no.3, pp. 482–493, 2003. [46] X. Gao, Q. Zhang, and Q. Tang, “Parachute dynamics and perturbation analysis of precision airdrop system,” Chinese Journal of Aeronautics,vol. 29, no.3, pp. 596–607,2016. [47] M. Clifford, “Detecting Freefall with Low-G Accelerometers,” Sensor and Analog Products Division Tempe, AZ, AN3151, Rev 0, 2006. International Journal of Advances in Rotating Machinery Multimedia Journal of The Scientific Journal of Engineering World Journal Sensors Hindawi Hindawi Publishing Corporation Hindawi Hindawi Hindawi Hindawi www.hindawi.com Volume 2018 http://www www.hindawi.com .hindawi.com V Volume 2018 olume 2013 www.hindawi.com Volume 2018 www.hindawi.com Volume 2018 www.hindawi.com Volume 2018 Journal of Control Science and Engineering Advances in Civil Engineering Hindawi Hindawi www.hindawi.com Volume 2018 www.hindawi.com Volume 2018 Submit your manuscripts at www.hindawi.com Journal of Journal of Electrical and Computer Robotics Engineering Hindawi Hindawi www.hindawi.com Volume 2018 www.hindawi.com Volume 2018 VLSI Design Advances in OptoElectronics International Journal of Modelling & Aerospace International Journal of Simulation Navigation and in Engineering Engineering Observation Hindawi Hindawi Hindawi Hindawi Volume 2018 Volume 2018 Hindawi www.hindawi.com Volume 2018 www.hindawi.com Volume 2018 www.hindawi.com www.hindawi.com www.hindawi.com Volume 2018 International Journal of Active and Passive International Journal of Antennas and Advances in Chemical Engineering Propagation Electronic Components Shock and Vibration Acoustics and Vibration Hindawi Hindawi Hindawi Hindawi Hindawi www.hindawi.com Volume 2018 www.hindawi.com Volume 2018 www.hindawi.com Volume 2018 www.hindawi.com Volume 2018 www.hindawi.com Volume 2018 http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Journal of Advanced Transportation Unpaywall

Design of Fully Automatic Drone Parachute System with Temperature Compensation Mechanism for Civilian and Military Applications

Loading next page...
 
/lp/unpaywall/design-of-fully-automatic-drone-parachute-system-with-temperature-FBJ3uyzacX

References

References for this paper are not available at this time. We will be adding them shortly, thank you for your patience.

Publisher
Unpaywall
ISSN
0197-6729
DOI
10.1155/2018/2964583
Publisher site
See Article on Publisher Site

Abstract

Hindawi Journal of Advanced Transportation Volume 2018, Article ID 2964583, 11 pages https://doi.org/10.1155/2018/2964583 Research Article Design of Fully Automatic Drone Parachute System with Temperature Compensation Mechanism for Civilian and Military Applications 1 2 2 Basem AL-Madani, Marius Svirskis, Gintautas Narvydas, 3 3 Rytis Maskeli0nas , and Robertas DamaševiIius Computer Engineering Department, College of Computer Science and Engineering, King Fahd University of Petroleum and Minerals, P.O. Box 1195, Dhahran 31261, Saudi Arabia Department of Automation, Kaunas University of Technology, StudentJ 48, 111, 51367 Kaunas, Lithuania Department of Multimedia Engineering, Kaunas University of Technology, Barsausko 59, A338, Kaunas, Lithuania Correspondence should be addressed to Robertas Damaˇsevici ˇ us; robertas.damasevicius@ktu.lt Received 23 May 2018; Accepted 4 November 2018; Published 15 November 2018 Academic Editor: Yair Wiseman Copyright © 2018 Basem AL-Madani et al. is Th is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. Application of Unmanned Aerial Vehicles (a.k.a. drones) is becoming more popular and their safety is becoming a serious concern. Due to high cost of top-end drones and requirements for secure landing, development of reliable drone recovery systems is a hot topic now. In this paper, we describe the development of a parachute system with fall detection based on accelerometer-gyroscope MPU – 6050 and fall detection algorithm based on the Kalman filter to reduce acceleration errors while drone is yfl ing. We developed the compensation algorithm for temperature-related accelerometer errors. The parachute system tests were performed from a small height on a soft surface. Later, the system was tested under real-world conditions. The system functioned effectively, resulting in parachute activation times of less than 0.5s. We also discuss the civilian and military applications of the developed recovery system in harsh (high temperature) environment. 1. Introduction be assigned to perform. Statistics of accidents during the operation of an Unmanned Aerial Vehicle (UAV) exhibits Unmanned Aerial Vehicles (UAVs), popularly known as a significantly higher accident rate compared to piloted drones, are autonomously or remotely operated aircrasft , aircrasft [16]. This poses severe limitations on the possible which can yfl without on-board human pilot, while being adoption of unmanned systems particularly in the civilian operated from the ground [1]. Currently, drones are for air space. During the flight, the drones may face a variety multiple purposes such as for entertainment (as a toy [2]), of problems, both internal (e.g., an electrical circuit failure, for civilian applications (for example, monitoring soil erosion broken connection, or mechanical damage) and external [3] and wildlife [4], remote sensing [5], surveying and pho- (interference of an external obstacle such as birds or hostile togrammetry [6], smart agriculture [7], disaster surveying force). Due to these problems, drones can become unman- and management [8], forestry [9], energy harvesting [10], ageable, fall down, and land on hard surface. In such cases, and housing renovation [11]), scientific applications [12], and not only the drone and its carrying equipment, but also military applications (surveillance and reconnaissance [13] the objects below (e.g., people or valuable property) can be and attacking the enemy [14, 15]). damaged resulting in large financial losses in addition to lost equipment and information. As the evolution of drone technology continues, its battery capacity and the range of the flight increase, therefore Hence, one important challenge is the implementation also increasing the range and scope of tasks drones can of technologies, which can deal gracefully with failures and 2 Journal of Advanced Transportation ensure safe operation even in the event of engine loss. In such as parachute, network, or hook-trapping. The descent ultralight aviation, emergency parachutes have become an of these systems can be automated using evolutionary algo- important tool to avert damages and casualties caused by rithms that help prepare the drone for landing, braking before aircraft failures. Parachute recovery is particularly suited to landing. tactical fixed wing UAV systems that require a high degree Recognition of a critical failure (due to internal mal- of mobility by allowing air vehicle recovery onto unprepared function or interaction with an external object or force) is terrain [17]. A higher attrition rate for UAVs compared to a distinctive problem in itself [27, 28]. Timely detection of manned aircraft can be due to both physical aircraft and critical failures can be used to enable activation of drone human operator factors. Given that the aircraft does not have recovery system in order to save an aircraft and/or sensitive to accommodate a human being, it can be made smaller. By data from being destroyed or falling into adversary hands. making the aircraft smaller, it is easier to stay undetected in Solutions include hardware redundancy to detect faulty sensors or equipment [29] and analytical redundancy using, the sky for longer time compared to a larger aircra.ft However, due to the size of the UAV, it will encounter dieff rent envi- e.g., fuzzy logic, to detect faulty sensor outputs [30], or ronmental issues that designers and operators may not have capturing an external view of an aircraft and subsequently had experience with in manned aircraft, thus requiring the applying image processing techniques to detect damage to the deployment of emergency recovery system such as parachute. body or parts of an aircraft [28]. However, due to the weight, Such problems could be avoided by using the drone- size, and cost limitations, it is not always possible to replicate mounted parachute system, which would expel parachute or put additional equipment on a small UAV such as a drone. during the fall and ensure safe landing. As in the course In case of critical failure determined, the autonomous of damage or hostile attack such as attempt of seizure, the parachute deployment systems are being the primary drone connection of drone to the control centre may be lost, recovery method now [31–33]. Autonomous Emergency sys- and itmay not bepossibleto activate theparachuteby tems are responsible for deploying the parachute in case of a sending a signal from the control centre. In addition, the critical failure (e.g., lost radio control signal or electrical or drone on-board sensors could detect a possible damage engine failure) [34]. The main advantage of the parachute- or loss of equipment during flight much faster and more based drone recovery method is that the UAV may be reliably, especially at high altitudes. Therefore, an automatic retrieved at any time anywhere (although an accurate point parachute system is a desired solution for the protection of landing is difficult). In addition, if a parachute is used the UAV [18]. It must also be reliable, to eliminate possible as the primary recovery method, it may also be used as mistakes and unintentional firing of the parachute [19], the emergency recovery method saving additional cost and because unnecessary parachuting at low altitude can damage weight. These systems are designed to safely locate aircrasft in objects on the ground. Avoiding such accidents requires emergency situations. Usually, the UAV ifl es autonomously to accurate identification of the free falling state regardless of a preselected site and the parachute is deployed ensuring safe theaxis and angleof thefall [20], but considering varying recovery. Most parachutes are parafoil or cruciform shaped, airflow the drone is subject to during the flight is a consider- stored in the fuselage. In cases where external sensitive able challenge. Analysing drone on-board sensor parameters equipment is under the fuselage, the parachute is deployed (such as acceleration [21]) can help to avoid false parachute from the belly and the vehicle is inverted so that the parts deployment, while it still ensures the protection of the drone under the fuselage are protected from the impact. A more from hitting the ground. complicated system for protecting the vehicle structure or any Operational and exploration requirements for UAVs sensitive parts is the combination of a parachute (deployed impose in particular cases the use of rescue and recovery normally on top of the fuselage) and an airbag which is systems. These systems could possess a large range of con- inflated under the fuselage a few seconds before touchdown. structive solutions, such as recover parachutes [22], airbags The airbag may alsobeused tokeep the UAV aofl at when [23], safety nets [24], and pneumatic cushions boats [25]. landed on water. The main problem in parachute recovery The rescue and recovery systems must allow a safe landing is the transition from horizontal to vertical motion of the in case of emergencies and could become a solution to a vehicle, which starts as soon as the parachute is deployed. standardization regarding landing in extreme weather and Then, although parachute steering is possible, the oscillatory geographical conditions [25]. movements of the vehicle present problems in recovery such The aim of this paper is to develop and experimentally as inaccurate point landing or touching down at different validate a drone-mounted parachute system capable of timely angles which may damage the vehicle structure [35]. parachuting and safely landing the drone in case of an Alizadeh et al. [23] suggested using airbag systems as accident, attack, or loss of control. a shock absorber for UAVs to assist with rapid parachute landings. An airbag system is activated on touchdown to absorb the kinetic energy during the impact with the ground. 2. State-of-the-Art Bleier et al. [34] presented the risk assessment models for generation of flight paths, which support the automatic Recovery of self-propelled yin fl g drones in a limited space is a rather difficult task [26]. During the halt, problems such deployment of emergency parachutes for Unmanned Aerial as hovering before landing, precise tracking of the UAV, Vehicles in emergencies due to loss of propulsion. Based on a risk analysis of the area underneath the flight path, and changing weather conditions are encountered. In this situation, there are several possible options for dropdowns suitable deployment positions are identified, which minimize Journal of Advanced Transportation 3 thechanceofendangering humans on theground, property 3. Method damage, and loss of the air vehicle. Du [36] introduced 3.1. Parachute Design Considerations. The purpose of a simulation model of a certain kind UAVs for UAV recovery parachute is to decrease speed while maintaining stability of system design. a falling vehicle or payload. The aerodynamic and stability Guo et al. [37] provided six degrees of freedom (DOF) characteristics of the parachute are determined by its geomet- flight dynamics model of UAV and parachute recovery ric features, which influence opening force and movement dynamic model for a quantitative prediction of the motion trajectory [43]. UAVs are subject to increasing demands and process of UAV and recovery system and predicted dangerous are committed to achieving maximum operational reliabil- situations at recovery stage. Kim et al. [31] predicted the ity. Parachutesystems are used to ensurethereliability of parachute deployment for landing at the desired point using operation and the safety of the machine. It is important to the neural network and the flight conditions such as the know the characteristics of the parachute for proper landing. deployment position, UAV’s velocity, and wind velocity as Parachutes can be of different types, and their parameters of input data sets, and landing points such as the cross range resistance to wind also vary greatly. Inflatable parachute is and the down range position as output data sets. Li et al. [32] suitable for slowing down at high altitude, where there is a analysed the motion characteristics of the parachute-UAV very rare atmosphere. Parachutes of this type are intended for combination in the whole recycling process and a comparison lowering scientific equipment and measure the direction of a of the calculated results to the experimental results to verify horizontal moving wind. Therefore, the release of a parachute the reliability of the numerical simulation. Shao et al. [22] must be fast and efficient. The effectiveness of the parachute evaluated a model of the UAV-parachute system, with wind spread depends on its size. A simple parachute, which does fields and a control strategy for recovery. not need to be controlled gradually, is effective only up to Shyu andHsiao [33] describedthe development of 1.5 m in diameter. Parachutes of this type are expanding mini-UAV that has the capabilities of bungee launch and very quickly; they can be used for discharging the opening parachute recovery. Yong and Li [38] studied the optimal container, from which the parachute simply falls out. Also, a parachute decelerating trajectory of mini-UAVs, which is spring exhaust mechanism can be used. solved by sequential quadratic programming method and Larger parachutes must withstand much higher overload genetic algorithm, while the results are used to guide due to high air resistance during take-off. In order to avoid research of parachute design. Zeng and Cai [39] introduced breaking of the cord, due to the occurrence of a shock, it is the parachute recovery system composed of fuselage, door, necessary to tighten the ropes before the parachute is pulled hanger, segregator, shooting system, main parachute, drogue out. One of the options is to use a rocket engine that pulls parachute, unlocking system of doors, and binding system. the parachute forward. However, such a system takes up quite The parachute system was optimized by wind tunnel tests a lot of space, which is difficult and not practical in small for the UAV being in unfavoured states by air-drop experi- aircra.ft Another option is to use a small parachute that drags ment. out a larger parachute. All variations must have the appro- Morgan et al. [40] explore the build and design of a priate fall speed during the launch of the parachute. When ballistic parachute recovery system. This system will monitor designing a parachute, several basic criteria need to be taken several variables in real-time to determine whether or not into consideration: simplicity of design and construction, the aircraft is operating in a safe environment. The elements resistance to tensile stress, and high wind resistance. observed are main battery voltage, current GPS coordinates, and current acceleration. If the system determines that the 3.2. Mathematical Model of Parachute Fall. It is difficult to aircraft is operating in an unsafe environment, the recovery model the dynamics of parachute ejection, opening, and system will cut main power and deploy a ballistic parachute falling. Usually some simplifications are adopted; for exam- to guide the aircraft safely to the ground. ple, the parachute is considered to be a rigid body which is Cristian and Codrea [41] designed parachute recovery affected by aerodynamic drag, strains the connection riser, and landing attenuation system for a military reconnaissance and transmits a force to the payload with six degrees of drone. The recovery system is able to recover the air vehicle freedom (see Figure 1). after the complete mission when the vehicle has landed in The system of equations presented here is based on the rough terrain at altitudes from sea level to 3000 m; recover previous worksof[43–46] andis denfi ed asfollows: the drone during uncontrolled flight conditions; and serve as safety device to prevent the air vehicle leaving the boundaries 𝑓 =𝑀 of the zone. Safe landing problems are also relevant in rocket systems 𝑓 =𝑋−𝑔𝑚 sin𝜃 [42]. Hybrid rockets can be used to control inflatable wings. (1) The advantage of such a system against a simple parachute 𝑓 =𝑍+𝑔𝑚 cos𝜃 system is that the landing is controlled and the inflatable 𝑓 =𝜃−𝛾−𝛼 wings help totravel tothe desired landing location. The inflatable wings occupy a much smaller area than the folds here, of similar dimensions. However, this type of wing system is difficult to reconcile, the wings are not very stable, they need 𝑀= 𝜌𝑉 𝑆 𝐶 +𝐹 cos𝛼+𝐹 sin𝛼 𝑝 𝑀 𝑅 𝑅 an inflated compressed gas control equipment. 𝑝 2 4 Journal of Advanced Transportation The angle of attack is given by 󵄨 󵄨 󵄨 󵄨 −1 󵄨 𝑧 󵄨 󵄨 󵄨 𝛼= tan (󵄨 󵄨 ) (7) 󵄨 󵄨 󵄨 󵄨 󵄨 󵄨 The angle of sideslip is derived by calculating - Tension Force 󵄨 󵄨 󵄨 󵄨 󵄨 𝑦 󵄨 −1 󵄨 󵄨 󵄨 󵄨 𝛽= sin ( ) (8) - Damping 󵄨 󵄨 󵄨 󵄨 󵄨 𝑉 󵄨 󵄨 󵄨 setting 3.3. Calculation of Drone Descent Speed. Drone speed is calculated in several steps. In order to know what acceleration the object moves, regardless of the accelerometer position, the sum of the acceleration values for all axes is expressed as 2 2 2 𝐺 =√𝐺 +𝐺 +𝐺 (9) 𝑥 𝑦 𝑧 where 𝐺 , 𝐺 , 𝐺 are the accelerometer values of the 𝑥 𝑦 𝑧 accelerometer axes (x, y, z). Assuming that v = 0 m / s, we can calculate the total fall speed as follows: Figure 1: Model of parachute system. 𝑡 𝑘𝑟 𝑘𝑟 󵄨 󵄨 →󳨀 󵄨 󵄨 󵄨 󵄨 (10) 𝑉 =∫ 󵄨 𝐺 󵄨 = Δ𝑡 ∑𝐺 𝑘𝑟 𝑧 󵄨 󵄨 2 󵄨 󵄨 𝑋= 𝜌𝑉 𝑆 𝐶 −𝐹 cos𝛼 𝑝 𝑝 𝑥 𝑅 where𝑡 is the time after which it is checked or the fall speed 𝑘𝑟 𝑍= 𝜌𝑉 𝑆 𝐶 −𝐹 sin𝛼 has not exceeded the set limit. 𝑝 𝑝 𝑧 𝑅 To eject the parachute emission, we apply the condition (2) that it must reach a certain speed (1.5 m/s) or more within a specified time interval (0.25 s). Otherwise, the parachute where M is the moment, X, Y,and Z are the body forces, 𝑉 will not be ejected. Under such conditions, the drone is not is velocity, S is parachute surface area, 𝐹 is the riser force allowed to reach a dangerous fall speed. (tension),𝛼 is attack angle, and𝛽 is sideslip angle. The deceleration of the parachute is calculated as follows: 3.4. Parameter Control. In order to understand the causes of 𝐹 −𝐹 −𝑚 𝑔 sin𝛾 𝑑𝑉 𝑅 𝐷 𝑝 a drone accident and to identify the distinctive features, it is (3) necessary to analyse previous accidents aiming to find possi- (𝑚 +𝑚 ) 𝑝 𝑑 ble solutions to problems. Having evaluated the real situations where 𝑚 is the parachute mass, 𝑚 is the drone mass, 𝐹 and in order for the algorithm to be able to distinguish the 𝑝 𝑑 𝑅 is riser force, 𝐹 is parachute drag force, and 𝛾 is flight path fall data from the movement data, the fall parameters must angle, and a is acceleration. be properly described. By measuring accelerometer readings The drag force of the parachute is calculated as follows: and evaluating parameters with a certain parameter limit, it is possible to detect the moment of falling. The system must clearly detect the drop and the rise 𝐹 = 𝑉 𝐷𝑆 (4) 𝐷 𝑝 regardless of the drone orientation in the space [47]. When knowing the exact moment from when the object began to where𝐷 is drag coefficient, 𝑆 is projected parachute area,𝑉 𝑝 𝑝 fall and when it reached the ground, it is possible to calculate is parachute velocity, 𝜌 is air density, and 𝜂 is an efficiency the height of the fall. Detection is also relevant to protecting factor. electronics from potential damage during impact. Detecting As a result, the following system of equations is con- a fall quickly is especially important in cases where the structed. operation has a very low height. This requires an optimized The falling speed is from the body velocities in three axes: program code and an appropriate microcontroller capable of 2 2 2 processing information as quickly as possible. 𝑉=√ 𝑉 +𝑉 +𝑉 (5) 𝑥 𝑦 𝑧 To determine the position and height change it is nec- essary to process the received sensor readings and to reject The deceleration is calculated as the derivative of the velocity incorrect values and calculate speeds at certain time intervals. vector: Position tracking algorithms with accelerometers can be 𝑉 𝑑𝑉 /𝑑𝑡+𝑉 𝑑𝑉 /𝑑𝑡+𝑉 𝑑𝑉 /𝑑𝑡 𝑑𝑉 𝑥 𝑥 𝑦 𝑦 𝑧 𝑧 used in situations where high precision is not required. The (6) 𝑉 integration algorithm is ideal for low-cost devices due to its Distance to impact Fall velocity & angle 𝑑𝑡 𝜂𝜌 𝑑𝑡 𝑥𝑦 𝑑𝑡 𝑥𝑦 Journal of Advanced Transportation 5 1.2 1.2 1.15 1.15 1.1 1.1 1.05 1.05 0.95 0.95 0.9 0.9 0.85 0.85 0.8 0.8 0 10 20304050 010 20 30 40 50 Temperature ( C) Temperature ( C) 2 ∘ Figure 2: Dependence of the accelerometer Z-axis values (1g – 9.81 m/s ) upon the temperature during rapid (5 C/min) heating (left) and cooling (right). simplicity. Also, for all speeds, all calculations are performed performed 10 times for each axis of the sensor, in a position using only oa fl ting point operations. The data is processed that maximizes the values of that axis. using a low frequency filter. Several different sensors can be A collection of values has been created in the processing used to increase the positioning accuracy. environment, which reads the values through a serial data output and writes them to a text file. The program reads 3.5. Calibration of Accelerometer. Accurate accelerometer sensor values and checks the conditions to avoid false values, values are particularly important for drone speed detection. checking whether the value arrives and whether it has more Due to temperature, accelerometer chip manufacturing inac- than 30 “char” elements. If conditions are met, values are curacies, and dieff rent stresses occurring during soldering written to the text document. This cycle is repeated until the and assembly, deviation of accelerometer values occurs. Mini- program is stopped (Figure 2). mal and maximum values of the scanned values are processed Figure 2 shows the deviations of the accelerometer val- by the Kalman lfi ter [48] (the delay in the submission of ues depending upon the temperature values. We can see data is not affected during the calibration). The filtered values that the acceleration values vary greatly. We assumed the are compared and recorded at the highest. Subsequently, the accelerometer can not warm up or cool down evenly due to subroutine is stopped, the accelerometer is overturned, the rapid heating or cooling (5 C/min). Therefore, it is necessary accelerometer axis being measured in the opposite direction, to significantly reduce the temperature variation in the and then the scan is continued. The same calibration steps are temperature control chamber. By reducing the temperature repeated for each axis. change to 1 C per minute, we obtained the following heating Accelerometers and gyros have high noise and temper- and cooling charts (Figure 3). ature variations. Therefore, it is not enough to take the From the graph (Figure 3) using linear regression, we values of the sensor data alone, but we also need to know have derived the following linear functions of the change the environmental conditions do the sensor values could be of acceleration during the cooling and heating process as compensated for temperature compensation. It is necessary follows: to measure the sensor values at different temperatures in 𝐺 = −0.0047𝑇+1.1031 order to determine possible errors. (11) A device for adjusting the temperature was made for 𝐺 = −0.0056𝑇+1.0967 calibration of the accelerometer. The thermoelectric heating and cooling element was used for this purpose, operating where𝐺 is the acceleration measured by the accelerometer’s under the Peltier effect principle. Also used are the temper- Z-axis. ature sensor DS18B20, the radiator for excessive heat or cold The initial and final values of the graphs are similar. of the thermoelectric element, a fan, a 12V power supply, Therefore, we can assume that during the cooling time the a bidirectional motor control TB6612FNG, and an Arduino accelerometer values need to be further elaborated. Since, microcontroller. by heating, the values change evenly without any significant By inserting the accelerometer into the temperature distortion, we assume that deviations of the Z-axis of the control chamber, rfi st of all, the chamber is positioned so sensor from the norm can be compensated by using this that the axis measured by the sensor is perpendicular to the function. The same test is performed with other axes (see the ground. The fan is mounted on the stand so that it does not results presented in Figures 4 and 5). touch the cooling system at all and does not aec ff t the sensor’s From Figures 4 and 5, we can note that the temperature- measurement data. Then, the sensor is cooled to 0 C, then induced variations on the X-axis and the Y-axis are con- ∘ ∘ heated to 50 C, and then cooled again to 0 C. The process is siderably smaller than the ones on the Z-axis. The Z-axis Acceleration (g) Acceleration (g) 6 Journal of Advanced Transportation 1.2 1.2 1.15 1.15 1.1 1.1 1.05 1.05 1 1 0.95 0.95 0.9 0.9 0.85 0.85 0.8 0.8 0 10 20304050 010 20 30 40 50 2 ∘ Figure 3: Dependence of the accelerometer Z-axis values (1g – 9.81 m/s ) upon the temperature during slow (1 C/min) heating (left) and cooling (right). 1.2 1.2 1.15 1.15 1.1 1.1 1.05 1.05 0.95 0.95 0.9 0.9 0.85 0.85 0.8 0.8 0 10 20304050 010 20 30 40 50 2 ∘ Figure 4: Dependence of the accelerometer X-axis values (1g – 9.81 m/s ) upon the temperature during slow (1 C/min) heating (left) and cooling (right). temperature compensation is derived from the heating value protocol between the receiver and the master controller. function, and the temperature compensation formula is as The main controller has several sensors: an accelerometer, a follows: magnetometer, a barometer, and a gyroscope. In order for the drone to yfl stably, it was necessary for this controller to 𝐺 =𝐺 +0.0047𝑇−0.1031 (12) 𝑧 𝑝𝑐𝑜𝑚 𝑧 adjust the PID parameters. The settings have been adjusted manually. where 𝐺 is the compensated acceleration value, 𝐺 is 𝑧 𝑝𝑐𝑜𝑚 𝑧 the measured value of the accelerometer Z-axis, and T is the 4.2. Electronics. A system for unlocking the parachute ejec- temperature. tion mechanism was created (Figure 8), which deploys the From the graphs of compensated values (Figure 6), we can spring holding rope. This system is particularly lightweight conclude that the deviations of the values are compensated. and has a small additional space, since it no longer requires However, due to differences in heating and cooling (at the a heavy servo motor. The Nichrome wire is used to burn same temperature), the values are compensated to a certain the cord. A MOS type transistor is used to prevent the cord limit, until the maximum deviation reaches 5%. The deviation from slipping off when the current is started. Transistor of the values before compensation was 14%. bandwidth is controlled by a PWM type signal. The system is managed by the Arduino microcontroller, which processes 4. Drone Design and Testing the sensor data, and controls the outputs. A battery of 110 4.1. Settings. Drone testing was performed under real condi- mAh lithium polymer has been selected for power supply. tions (see Figure 7). The drone frame consisted of a balsam Its charging system has an integrated battery charger with tree reinforced with a glass fibre. This type of frame is battery protection from excessive discharge. The battery is charged through the micro USB connector. The RGB LED extremely lightweight and has low production and repair costs. The electric motor bearing structures consisted of is used for indicating the operation. The LED at the start of aluminium hollow bars. The Emax company controller was the system, in colour, indicates the level of the battery voltage selected for drone engine management. The drone is con- and informs about the operation of the system and the fall trolled using the 9-channel receiver using the S-Bus type detection. Journal of Advanced Transportation 7 1.2 1.2 1.15 1.15 1.1 1.1 1.05 1.05 0.95 0.95 0.9 0.9 0.85 0.85 0.8 0.8 010 20 30 40 50 0 10 20304050 2 ∘ Figure 5: Dependence of the accelerometer Y-axis values (1g – 9.81 m/s ) upon the temperature during slow (1 C/min) heating (left) and cooling (right). 1.2 1.2 1.15 1.15 1.1 1.1 1.05 1.05 1 1 0.95 0.95 0.9 0.9 0.85 0.85 0 10 20304050 0 10 20304050 2 ∘ Figure 6: Dependence of the compensated accelerometer Z-axis values (1g – 9.81 m/s ) upon the temperature during slow (1 C/min) heating (left) and cooling (right). Figure 8: Parachute system electronic circuit diagram: 1 – charging unit; 2 - switch; 3 – Lithium polymer battery; 4 - microcontroller; 5 – accelerometer - gyroscope; 6 – LED diode; 7 – MOS type transistor; Figure 7: Ready-to-yfl drone. 8 – nichrome wire. 4.3. Control Algorithm. For fall detection, we have adopted an approach that is based on the detection of the falling state axis values are lfi tered using a Kalman filter. Using the filtered of the drone. The approach in itself in not new [49] and we data, the combined acceleration of all axles is calculated. selected it due to its simplicity, however, we have improved Then, the condition is checked whether the speed limit is the approach by introducing compensation for temperature xfi ed. If the limit is xfi ed - the time from the moment of caused variability. The parachute system program (Figure 9) capture is measured. After a limited time, the fall acceleration measures the battery voltage, then the accelerometer and is checked again. If the condition ts, fi the parachute is ejected. gyroscope data. The Z-axis data on the accelerometer is Also, one of the main conditions for detecting the fall state or compensated by the temperature-dependent deviation. The the unstable operation of a drone is the measurement of the 8 Journal of Advanced Transportation START Start calculating time, set first variable to 1 No Yes Is voltage Initiate sensor >3.8V? readout Read accel. data Calculate acceleration on all axis Is Yes No Initiate angle accel. Value Is this an initial equal to fall accel. Readout? readout value? No Yes Read angle data No Yes Is angle >90 Drop the parachute Yes No End of First variable is Deadlock on? programme marked as 0 Figure 9: Parachute system operation algorithm. angle of inclination. When the limit value (at 90 ) is exceeded, (the system was thrown onto a soft surface). First of all, the parachute is ejected. If no condition is satisfied, the cycle the simple fall detection algorithm was tested: when the is repeated until the condition is satisfied or the shutdown average of the total values of the acceleration was 15 and button is depressed. the threshold was applied to it. When crossing the limit, the parachute system was triggered. The system worked 4.4. Parachute Exhaust System. The parachute was made reliably. However, when placed on the drone, no fall was from a rugged nylon material. Parachute dimensions are recorded due to excessive vibrations. Using the Kalman filter for this algorithm, the system worked efficiently. The fall chosen such that a 2 kg drone does not fall faster than 5 m/s. Under the selected dimensions, a 16-segment dome-shaped was recorded in 0.5 s. In order to prevent the unintentional parachute with overlock stitches was made (Figure 10). The ejection, the fall acceleration was checked for a second time diameter of the parachute is 1.45 m. aer ft a specified period of time. Only in accordance with The parachute exhaust housings (Figure 11) were made the second measurement condition the parachute is to be of PVC pipe, glass br fi e board, and a bottom-mounted opened. During critical conditions, when vibrations have mounting piece made of three-dimensional printing. extremely high amplitude and frequency, both Kalman and the compliant filter failed to properly lfi ter the errors. As a result, the fall was sometimes erroneously detected. However, 4.5. Testing. To determine the efficiency of the parachute system, we attempted to detect a fall without the drone itself in real conditions, we have not detected any false falls. The Journal of Advanced Transportation 9 Figure 12: System testing under real conditions. Figure 10: eTh 16-segment, dome-shaped parachute. of different environments, while ensuring the avoidance of obstacles [50]. For example, the meteorology drones collect data on air temperature, humidity, pressure, wind force, radi- ation, etc. and should beabletowithstand theadverse forces of nature such as strong wind gusts. If yfl ing at high altitudes also the temperature factor should be taken into account [51]. There are significant challenges related to predicting tem- perature variations when using drones in Arctic [52], where temperature may fall below -50 C, or in deserts [53], where Figure 11: Parachute system body. temperature may exceed +50 C, but also for more common civilian applications executed during nigh time [54], or for forest fire monitoring [55], when temperatures may fall below operating parachute system during opening stage is depicted or well exceed the range of temperatures most consumer ∘ ∘ in Figure 12. drones are designed for, which is usually 0 C–40 C. In During the tests, an attempt was made to accelerate the this paper we have addressed the challenge of designing a parachute opening, and the parachute was covered with Talc reliable parachute-based recovery system for a drone, which powder from the bonding of the material. A special parachute uses the temperature-compensated accelerometer data for folding technique is used when the parachute is folded in parachute ejection. Note that still many other temperature- such a way as to spill as soon as possible. As a result, folded related challenges for drone designers exist, most notably, parachute can open much faster. However, the minimum the impact of temperature on the battery performance and height required for the parachute to land was about 20 m from characteristics. the ground. 6. Conclusion 5. Evaluation We have proposed a parachute-based system for drone recov- We evaluate the developed UAV parachute system according ery with the accelerator sensor temperature compensation to the requirements for UAV recovery systems formulated by mechanism. The parachute system has been successfully Wyllie [17] as follows: tested in real-life conditions. The fall is detected almost immediately (within 0.5 s). However, it takes much more time (i) Safety: a mandatory requirement during landing. for the parachute to unfold, so at less than 20 m height, the (ii) Protection: protect drone and sensitive on board parachute still may not be able to unfold in time. The system equipment from damage during landing. also includes a mechanism for prevention of unintended parachute deployment. In calculating the rate of fall, the (iii) Accuracy: ensure high accuracy when landing to a resulting acceleration errors, even aeft r compensation, were planned landing point to minimize landing damage. high. The calculated speed data could not be relied upon even (iv) Automation: reduce the number of manual opera- at small intervals, assuming that an initial speed of falling is tions by the drone operator. equal to 0 m/s. The acceleration deviations can be offset by (v) Reliability: ensure predictability of operation. applying the Kalman lfi ter. The accelerometer values depend significantly upon the (vi) Repeatability: ensure that recovery may be repeated environment temperature, so the compensation mechanism many times during lifetime of the drone. based on linear regression was introduced to allow stable Since the areas where drones are used are expanding with new detection of drone fall state during harsh environment con- applications proposed every year, it is important that drones ditions, such as in a desert environment during a hot sum- continue to be used both safely and efficiently in multitude mer. 10 Journal of Advanced Transportation Data Availability systems (UASs), part 2: scientific and commercial applications,” Journal of Unmanned Vehicle Systems, vol.02,no.03,pp. 86–102, The data used to support the findings of this study are available from the corresponding author upon request. [13] C. Paucar,L.Morales, K.Pinto et al., “Use of drones forsurveil- lance and reconnaissance of military areas,” in Developments and Advances in Defense and Security. MICRADS 2018. Smart Conflicts of Interest Innovation, Systems and Technologies, A. Rocha and T. Guarda, Eds., vol.94, pp.119–132,Springer, Cham,2018. The authors declare that they have no conflicts of interest. [14] J. Drew, “Drone strike: Long-range attack UAVs being devel- oped from aerial targets,” Aviation Week & Space Technology, Acknowledgments vol.179, no.4,pp.42-43,2017. The authors acknowledge the support by King Fahd Univer- [15] D. Hambling, “Military tackles the problem and the potential of drones,” Aviation Week & Space Technology, vol. 180, no. 8, pp. sity of Petroleum & Minerals (KFUPM) in publishing this 44–47, 2018. work. [16] G. Wild, J. Murray, and G. Baxter, “Exploring Civil Drone Acci- dents and Incidents to Help Prevent Potential Air Disasters,” References Aerospace, vol. 3, no. 3, p. 22, 2016. [1] E. N. Barmpounakis, E. I. Vlahogianni, and J. C. Golias, “Un- [17] T. Wyllie, “Parachute recovery for UAV systems,” Aircraft Engineering and Aerospace Technology, vol.73, no.6,pp.542– manned Aerial Aircraft Systems for transportation engineering: Current practice and future challenges,” International Journal of 551, 2001. Transportation Science and Technology, vol. 5,no. 3, pp.111–122, [18] K. Graham and S. Cartwright, Feasibility of Parachute Recovery Systems for Small UAVs, University of New South Wales, New South Wales, 2008. [2] C.-H. Poh and C.-K. Poh, “Radio Controlled “3D Aerobatic Air- planes” as Basis for Fixed-Wing UAVs with VTOL Capability,” [19] D. Lim,C. Park, N. H.Kim,S.-H. Kim,and Y. S.Yu, “Fall- Detection Algorithm Using 3-Axis Acceleration: Combination Open Journal of Applied Sciences,vol.04,no. 12,pp.515–521, 2014. with Simple Threshold and Hidden Markov Model,” Journal of Applied Mathematics,vol.2014,ArticleID896030,8pages,2014. [3] S. D’Oleire-Oltmanns, I. Marzolff, K. D. Peter, and J. B. Ries, [20] J. Luo, B. Zhong, and D. Lv, “Fall Monitoring Device for Old “Unmanned aerial vehicle (UAV) for monitoring soil erosion People based on Tri-Axial Accelerometer,” International Journal in Morocco,” Remote Sensing, vol. 4, no. 11, pp. 3390–3416, 2012. of Advanced Computer Science and Applications, vol.6,2015. [4]M.E.Goebel, W. L. Perryman, J.T. Hinke et al., “A small [21] K. Seifert, “Oscar Camacho. Implementing Positioning Algo- unmanned aerial system for estimating abundance and size of rithms Using Accelerometers,” Application Note AN3397, Rev 0, Antarctic predators,” Polar Biology,vol.38,no.5,pp.619–630, Freescale Semiconductor,2007. [22] P. Shao, C.Wu, and S.Ma, “Research onkey problems in [5] K. Whitehead and C. H. Hugenholtz, “Remote sensing of the assigned-point recovery of UAV using parachute,” in Proceed- environment with small unmanned aircraft systems (UASs), ings of the 2013 IEEE International Conference of IEEE Region 10 part 1: a review of progress and challenges,” Journal of Un- (TENCON 2013),pp. 1–4, 2013. manned Vehicle Systems,vol. 02, p. 69,2014. [23] M. Alizadeh, A. Sedaghat, and E. Kargar, “Shape and orifice [6] F.Cuesta, F. Lopez-Rodriguez, and A.Esteban,“A New Blondin optimization of airbag systems for UAV parachute landing,” System for Surveying and Photogrammetry,” Sensors,vol.13,no. International Journal of Aeronautical and Space Sciences,vol. 15, 12, pp. 16894–16914, 2013. no. 3, pp. 335–343, 2014. [7] E. Honkavaara, H. Saari, J. Kaivosoja et al., “Processing and [24] S. Ramasamy, R. Sabatini, and A. Gardi, “A Unified Analytical assessment of spectrometric, stereoscopic imagery collected Framework for Aircraft Separation Assurance and UAS Sense- using a lightweight UAV spectral camera for precision agricul- and-Avoid,” Journal of Intelligent & Robotic Systems,vol.91,no. ture,” Remote Sensing, vol.5,no. 10,pp.5006–5039,2013. 3-4, pp. 735–754, 2018. [8] Z.Xu, J. Yang, C.Peng etal.,“Development ofanUAS for [25] V. Prisacariu, “Recovery System Of The Multi-Helicopter Uav,” post-earthquake disaster surveying and its application in Ms7.0 Review of the Air Force Academy, vol.1, no. 31,pp.91–98, 2016. Lushan Earthquake, Sichuan, China,” Computers & Geosciences, [26] S. Khantsis, Control system design using evolutionary algorithms vol. 68, p. 22, 2014. for autonomous shipboard recovery of unmanned aerial vehicles. [9] L. Tang and G. Shao, “Drone remote sensing for forestry Doctor of Philosophy (PhD), Aerospace, Mechanical and Manu- research and practices,” Journal of Forestry Research,vol. 26, no. facturing Engineering, RMIT University, 2006. 4, pp. 791–797, 2015. [27] C. H.Lo, E. H.Fung, and Y.K.Wong, “Intelligent automatic [10] Z. Zhao, X. Pu, C. Du et al., “Freestanding Flag-Type Triboelec- fault detection for actuator failures in aircra,ft” IEEE Transac- tric Nanogenerator for Harvesting High-Altitude Wind Energy tions on Industrial Informatics, vol.5,no. 1,pp. 50–55, 2009. from Arbitrary Directions,” ACS Nano,vol.10, no.2,pp. 1780– [28] Y. Wiseman, “Device for detection of fuselage defective parts,” 1787, 2016. International Information Institute (Tokyo). Information,vol. 17, [11] A. A. Mustaffa, M. F. Hasmori, A. S. Sarif, N. F. Ahmad, and N. Y. no. 9, pp. 4189–4194, 2014. Zainun, “The Use of UAV in Housing Renovation Identification: [29] D. Rehage, U. B. Carl, and A. Vahl, “Redundancy manage- A Case Study at Taman Manis 2,” IOP Conference Series: Earth ment of fault tolerant aircraft system architectures - Reliability and Environmental Science,vol.140,p.012003,2018. synthesis and analysis of degraded system states,” Journal of [12] K. Whitehead, C. H. Hugenholtz, S. Myshak et al., “Remote Aerospace Science and Technology, vol. 9, no. 4, pp. 337–347, sensing of the environment with small unmanned aircraft 2005. Journal of Advanced Transportation 11 [30] W. A. Kwong, K. M. Passino, E. G. Laukonen, and S. Yurkovich, [48] E. A. Wan and R. Van Der Merwe, “eTh unscented Kalman “Expert supervision of fuzzy learning systems for fault tolerance filter for nonlinear estimation,” in Proceedings of the IEEE Adap- aircraft control,” Proceedings of the IEEE, vol. 83, no.3,pp. 466– tive Systems for Signal Processing, Communications, and Con- 482. trol Symposium (AS-SPCC 2000), pp. 153–158, IEEE, Alberta, Canada, October 2000. [31] I. Kim, S. Park,W.Park,and C.-K.Ryoo, “Prediction of the parachute deploy for landing at the desired point,” in Proceed- [49] K. L. Crandall and M. A. Minor, “UAV fall detection from a dynamic perch using Instantaneous Centers of Rotation and ings of the 2012 Proceedings of SICE Annual Conference, SICE 2012, pp. 1629–1634, IEEE, Akita, Japan, August 2012. inertial sensing,” in Proceedings of the 2015 IEEE International Conference on Robotics and Automation, ICRA 2015,pp.4675– [32] M. Li, H. Zhang, and J. Chen, “Dynamic analysis for parachute 679,USA,2015. recovery system of unmanned aerial vehicle,” Journal of System [50] A. Ivanovas, A. Ostreika, R. Maskeliu¯nas, R. Damaˇsevici ˇ us, Simulation,vol.24, no.6,pp.1303–1307,2012. D. Połap, and M. Wozniak, “Block Matching Based Obstacle [33] L.-S. Shyu and Y.-C. Hsiao, “Mini UAV design and manufacture Avoidance for Unmanned Aerial Vehicle,” in Artificial Intelli- with bungee launched / Parachute recovery,” Applied Mechanics gence and Soft Computing , pp. 58–69, Springer International and Materials,vol.610, pp.97–100, 2014. Publishing, 2018. [34] M. Bleier, F. Settele, M. Krauss, A. Knoll, and K. Schilling, “Risk [51] E. H.Teets Jr., C.J.Donohue,K. Underwood,and J.E. Bauer, assessment of flight paths for automatic emergency parachute “Atmospheric considerations for uninhabited aerial vehicle deployment in UAVs,” IFAC-PapersOnLine, vol.28, no.9,pp. (UAV) flight test planning,” NASA/TM-1998-206541, 1998. 180–185, 2015. [52] M. Ader and D. Axelsson, Drones in Arctic Environments,KTH, [35] W. J. Crowther and K. Prassas, “Post Stall Landing for Field Stockholm, Sweden, 2017. Retrieval of UAVs,” in Proceedings of the 14th Bristol Interna- [53] D. Gallacher, “Applications of micro-UAVs (drones) for desert tional unmanned air vehicle systems conference, 1999. monitoring: current capabilities and requirements,” in Proceed- [36] K. M. Du, “Simulation analysis on the problem of opening main ings of the Kubuqi International Desert Forum 2015,pp. 249–259, parachute used in UAV recovery system,” Applied Mechanics and Materials,vol.390,pp. 57–61,2013. [54] S. R. Herwitz, K. Allmendinger, R. Slye et al., “Nighttime UAV [37] L. Guo, H. Zhang, and M. Tong, “Dynamics analysis on Vineyard Mission: Challenges of See-and-Avoid in the NAS,” parachute recovery of unmanned aerial vehicle,” Journal of in Proceedings of the AIAA 3rd “Unmanned Unlimited” Techni- Nanjing University of Aeronautics and Astronautics,vol. 44,no. cal Conference, Workshop and Exhibit, Chicago, Illinois, USA, 1, pp. 14–19, 2012. September 2004. [38] Z. Yong and L. Li, “Parachute Decelerate Trajectory Optimiza- [55] P. Skorput, S. Mandzuka, and H. Vojvodic, “The use of Un- tion Design Based on Genetic Algorithm,” in Proceedings of the manned Aerial Vehicles for forest fire monitoring,” in Proceed- 2009 International Workshop on Intelligent Systems and Appli- ings of the 58th International Symposium ELMAR, ELMAR 2016, cations, ISA 2009, IEEE, Wuhan, China, 2009. pp. 93–96, Croatia, September 2016. [39] L. Zeng and J. Q. Cai, “Scheme design of a parachute recovery system for UAV,” in Proceedings of the 64th International Astro- nautical Congress 2013, IAC 2013, pp. 8404–8412, China, 2013. [40] A. Morgan and R. Chapman, Ballistic Parachute Recovery Sys- tem for Unmanned Aerial Vehicles (UAVs), Make Magazine, [41] A.-I. Cristian and S. Codrea, “Parachute deployment systems for drones, applied on uav-sact project,” in 15th International Conference eTh Knowledge-Based Organization: Applied Techni- cal Sciences and Advanced Military Technologies,vol.6,pp.260– 266, 2009. [42] C. Sudduch, Design of a Hybrid Rocket/Ina fl table Wing UAV , Oklahoma State University, USA, 2012. [43] G. Guglieri, “Parachute-Payload System Flight Dynamics and Trajectory Simulation,” International Journal of Aerospace Engi- neering, vol. 2012, Article ID 182907, 17 pages, 2012. [44] K.-F. Doherr and H. Schilling, “Nine-degree-of-freedom simu- lation of rotating parachute systems,” Journal of Aircraft ,vol.29, no.5,pp.774–781,1992. [45] V. N.Dobrokhodov,O.A. Yakimenko, and C.J.Junge,“Six- degree-of-freedom model of a controlled circular parachute,” Journal of Aircraft ,vol. 40, no.3, pp. 482–493, 2003. [46] X. Gao, Q. Zhang, and Q. Tang, “Parachute dynamics and perturbation analysis of precision airdrop system,” Chinese Journal of Aeronautics,vol. 29, no.3, pp. 596–607,2016. [47] M. Clifford, “Detecting Freefall with Low-G Accelerometers,” Sensor and Analog Products Division Tempe, AZ, AN3151, Rev 0, 2006. International Journal of Advances in Rotating Machinery Multimedia Journal of The Scientific Journal of Engineering World Journal Sensors Hindawi Hindawi Publishing Corporation Hindawi Hindawi Hindawi Hindawi www.hindawi.com Volume 2018 http://www www.hindawi.com .hindawi.com V Volume 2018 olume 2013 www.hindawi.com Volume 2018 www.hindawi.com Volume 2018 www.hindawi.com Volume 2018 Journal of Control Science and Engineering Advances in Civil Engineering Hindawi Hindawi www.hindawi.com Volume 2018 www.hindawi.com Volume 2018 Submit your manuscripts at www.hindawi.com Journal of Journal of Electrical and Computer Robotics Engineering Hindawi Hindawi www.hindawi.com Volume 2018 www.hindawi.com Volume 2018 VLSI Design Advances in OptoElectronics International Journal of Modelling & Aerospace International Journal of Simulation Navigation and in Engineering Engineering Observation Hindawi Hindawi Hindawi Hindawi Volume 2018 Volume 2018 Hindawi www.hindawi.com Volume 2018 www.hindawi.com Volume 2018 www.hindawi.com www.hindawi.com www.hindawi.com Volume 2018 International Journal of Active and Passive International Journal of Antennas and Advances in Chemical Engineering Propagation Electronic Components Shock and Vibration Acoustics and Vibration Hindawi Hindawi Hindawi Hindawi Hindawi www.hindawi.com Volume 2018 www.hindawi.com Volume 2018 www.hindawi.com Volume 2018 www.hindawi.com Volume 2018 www.hindawi.com Volume 2018

Journal

Journal of Advanced TransportationUnpaywall

Published: Nov 15, 2018

There are no references for this article.