Design and Integration of a Drone based Passive Manipulator for Capturing Flying Targets
B. V. Vidyadhara, Lima Agnel Tony, Mohitvishnu S. Gadde, Shuvrangshu Jana, V. P. Varun, Aashay Anil Bhise, Suresh Sundaram, Debasish Ghose
RRobotica (2021) © Cambridge University Press 2021doi:
Design and Integration of a Drone based PassiveManipulator for Capturing Flying Targets
Vidyadhara B. V. ∗ , Lima Agnel Tony † , Mohitvishnu S.Gadde , Shuvrangshu Jana , Varun V. P. , Aashay AnilBhise , Suresh Sundaram , Debasish Ghose Guidance, Control, and Decision Systems Laboratory (GCDSL), Department ofAerospace Engineering, Indian Institute of Science, Bangalore-12, India. Robert Bosch Center for Cyber Physical Systems, Bangalore-12, India. Artificial Intelligence and Robotics Laboratory (AIRL), Department of AerospaceEngineering, Indian Institute of Science, Bangalore-12, India.Emails: [email protected], [email protected], [email protected],[email protected], [email protected], [email protected] (Accepted MONTH DAY, YEAR. First published online: MONTH DAY, YEAR)
SUMMARY
In this paper, we present a novel passive single Degree-of-Freedom (DoF) manipulatordesign and its integration on an autonomous drone to capture a moving target. The end-effector is designed to be passive, to disengage the moving target from a flying UAV andcapture it efficiently in the presence of disturbances, with minimal energy usage. It isalso designed to handle target sway and the effect of downwash. The passive manipulatoris integrated with the drone through a single Degree of Freedom (DoF) arm, andexperiments are carried out in an outdoor environment. The rack-and-pinion mechanismincorporated for this manipulator ensures safety by extending the manipulator beyondthe body of the drone to capture the target. The autonomous capturing experiments areconducted using a red ball hanging from a stationary drone and subsequently from amoving drone. The experiments show that the manipulator captures the target with asuccess rate of 70% even under environmental/measurement uncertainties and errors.KEYWORDS: Aerial manipulation; Passive end-effector; Moving target capture.
1. Introduction
Technological advancements in Unmanned Aerial Vehicles (UAVs) have led to the growthof industries developing solutions for various civilian and military applications. UAVsare extensively used for various applications like aerial photography, package delivery,mapping of difficult terrain or environments, reforestation, visual inspection, search andrescue operations, etc. Drone delivery networks like Amazon Prime Air, UAV basedmedical transportation like Zipline, drone photography with DJI, Skydio are platformsof a few popular application. Due to their complexity, UAVs and their related sub-systemspose a challenge to researchers. Recently, aerial manipulation is gaining attention due toits wide scope for applications. Manipulation mechanisms are key to robotics applications.An ideal manipulator should be energy optimal and have low response time and structuralintegrity. Fields in which this domain will have an impact include material handling, ∗ Corresponding author. E-mail: [email protected] † Corresponding author. E-mail: [email protected] a r X i v : . [ c s . R O ] F e b Design and Integration of a Drone based Passive Manipulator for Capturing Flying Targets inventory management, package delivery, etc. Steps towards achieving similar tasks areavailable in the literature.The literature has diverse works on aerial manipulation
6, 7 for different applications.Controlling a valve using multiple DoF manipulator and adopting parallel manipulatorsfor turning are those which are suitable for localised manipulation with higher accuracy.Tri-finger end effector design is adopted in, which is a foldable one and utilises lesspower. Manipulation using an industrial manipulator on a helicopter is also adoptedwhere, a seven DoF manipulator is employed to study the coupling effects of theintegrated system. This manipulator has better maneuverability but the end-effectoroperational area is small with limited reach and considerable weight. Only a few researchpapers
12, 13 in the literature discuss the design and modeling of aerial manipulators. Pickand place operation using haptic control of manipulators are looked into, where theend-effectors are inefficient for a dynamic task and have limited operational volume.Manipulators for cooperative transportation of objects work together to achievestatic object transportation. Manipulators for contact-based operations are also activelyresearched
18, 19 for problems like pipeline and power line monitoring and repair andsimilar applications. Aerial grasping of objects is another interesting area. A vision-based
Target drone with the ball
Drone with manipulator arm
Fig. 1: Sample scenario describing the problemgrasping is presented in, where a multi-degree of freedom robotic arm is considered.Aerial manipulation of a rod-shaped object using multiple robots is another stationaryobject manipulation in which, the manipulator’s task is to grip the longitudinally placedobject. Grasping of cylindrical objects is presented in while a suction-based end effectorfor pick and place is given in.
23, 24
An extending zipper manipulator for aerial graspingis presented in, while a seven DoF manipulator with a hex-rotor for object grasping.The above-mentioned designs are used to interact with static targets. They mainlyincorporate multi DoF robotic arm concepts for the task. These designs are heavy, havehigh power consumption and add computational and mechanical complexity. For theconsidered task, which is extremely dynamic and fast, these are major issues. While thedownwash and vibration aspects are not serious concerns in the reviewed literature, it isa serious problem for a dynamic target capture task. Most of the works that deal withmoving object grasping are dealt with in a control perspective rather than a manipulatordesign perspective. Hence, a novel design that can interact with dynamic targets andaccommodate considerable sensor information error is a much-needed research and hasapplications in several domains.In this paper, we present modelling and development of a single DoF manipulatorfor aerial grabbing of stationary and moving objects in an outdoor environment. Thedesign contributes to low drag and low impact from downwash. The proposed passiveend-effector design is energy optimal and any object within 0.15 kg and 0.2 m diametercould be grabbed. The design of the manipulator is presented along with the analysison the manipulator modelling parameters and the stability of the integrated system.The paper also presents results demonstrating the performance of the manipulator whilegrabbing stationary and moving object. esign and Integration of a Drone based Passive Manipulator for Capturing Flying Targets
2. Problem Description
The manipulator is designed for aerial grabbing of moving targets. The problemrepresented in Fig. 1, is inspired by Challenge 1 of MBZIRC 2020, in which a dronecarries a ball attached to it with a flexible rod. The drone moves with a maximum speedof 6 m/s, and the ball weighs 0 .
060 kg and is 0 .
15 m in diameter. The contact betweenthe rod and the ball is magnetic. The task is considered successful if the drone can detachthe ball and drop the captured ball in a box. The ball is prone to oscillations becauseof the drone maneuvers and environmental factors like wind gusts and downwash. Thedesign requirements for the manipulation mechanism are listed below.a. The volume of the integrated system should be within 1 . × . × .
3. Manipulator Design Challenges
The design of the manipulator and the end-effector should address these challenges foreffective grasping.a. Location of the manipulator: Selecting the location to mount the manipulator iscrucial. Fig. 2 represents the possible location of end-effector. Location 1 has a largeusable volume. Location 1 and location 3 have the advantage that they can be placedvery close to the center of drone’s frame but pose the risk of the drone or the ballstriking the propellers. Metallic construction around the GPS can cause problems,making location 1 a risky choice. Locations 2 and 4 have large usable volumes butare affected by the downwash from the propellers. Locations 2, 3, and 4 require themanipulator to extend away from the drone body to ensure safety and avoid downwash.It may generate moments about the drone centre of gravity (CG) if it is not properlystabilised.
Design and Integration of a Drone based Passive Manipulator for Capturing Flying Targets
AB CD
Fig. 3: Manipulator arm design considerationsFig. 4: Requirement of a force to remove targetb. Length of the extension arm: The end-effector should be at a safe distance from thepropellers, to ensure safety. This is done by extending the end-effector away from thedrone body, as shown in Fig. 3 using different mechanisms. It could be a single DoF(A/B), two DoF (C) or multiple DoF (D) manipulation mechanisms. Multiple DoFimproves the reachable space of the manipulator at the cost of increased computationaland control complexity. Having multiple DoF adds to the reachable space of themanipulator but at the cost of increased computational and control complexity. Whilea considerable extension of the arm is recommended for the safety of the drone, itmight contribute to several other issues. Large distances from the point of attachmentcreate a noticeable deflection at the end-effector. For multiple DoF joints, this affectsthe performance of actuators resulting in sensor errors due to change in orientationand position. Such extensions also create moments that tend to destabilize the drone.c. Detachment force: The forces encountered while detaching a ball from the target UAVare shown in Fig. 4. The ball is attached to the rod suspended from the UAV airframe, by a magnet. The detachment force (denoted by F d in Fig. 4) is non-uniformas it depends on the way that the magnets are pulled apart, and thus depends on themechanism of grasping. In addition, the target UAV is moving at a velocity v , whichcauses a force on the end-effector as a result of the kinetic energy acquired by theball. The end-effector must be able to exert the maximum necessary force. Creating arobust manipulator with a high factor of safety, results in an increase in weight addingto the moment and sag issues. This challenge relies on strength to weight optimisationand material selection.d. Vibrations: The end-effector self-weight causes deflection, as mentioned above. Thisdeflection imparts vibrations. The manipulator along with the end-effector acts as anend loaded cantilever beam which is prone to vibrations even for small disturbancesat the free end. Hence, sensor vibration dampening becomes crucial for this task, toremove noise from the vision feed. esign and Integration of a Drone based Passive Manipulator for Capturing Flying Targets Detachment points Detachment points
Fig. 5: (a) CAD model of the final passive ball grabbing end-effector (b) A workingprototype
Limit switches
Thin plate for detection (a) (b)
Fig. 6: Grab detector (a) CAD model (b) Prototype (coin for scale)
4. Passive Manipulator Design Approach
The challenges involved in the problem are thoroughly examined and the following designis proposed. The design requires certain essential capabilities like quick response, lowweight, and optimal power consumption. The design is a result of iterative and progressivedevelopments from a preliminary concept. The motivation of the manipulator mechanism comes from the passive fruit pickers usedin orchards. The design is as shown in Fig. 5(a) and the prototype of this design is shownin Fig. 5(b). As seen in the figures, the top portion of the end effector has a sinusoidalshape made from birch, with several detachment points made from carbon fibre (CF)strips. A hand woven nylon mesh is attached to its bottom and is supported at the frontusing a semi-circular CF ring. The specific shape of the top not only supports the meshbelow but also improves its effectiveness. The top portion of the end-effector is convex atits center and concave towards the ends. The center portion of the end-effector has 3 Dprinted mounts to place the camera (eye-in-hand configuration) using CF tubes and toattach the basket on to the manipulator arm. The convexity ensures that the ball remainswithin the FoV of the camera until it is detached, which otherwise would happen to theside or behind the camera. The concave shape towards the end give sufficient room for theball to be collected in the basket. If not, the ball may fall outside while being detachedfrom the drone. The CF detachment points aid in effectively detaching the ball.Since, the target needs to be dropped in a box, a dropping mechanism is integratedinto the system. The servo motor present in the dropping mechanism helps in releasingthe ball in the box. The servo motor requires lesser power to operate satisfying theminimal power requirement. An additional requirement for the manipulator is to detectthe grabbed ball. Approaches like visual feedback with a separate camera or gimbalmounted camera would add computational load on the system, with considerable energyrequirement. Thus, a thin plate detector is designed with three switches placed arounda circle, as shown in Fig. 6(a). The plate at its center improves sensitivity and detectsthe grabbed ball. The design uses gravity for grab detection and release of the ball intothe basket, with minimal use of energy. The prototype of the mechanism, is shown inFig. 6(b). The gray rim is held firmly by the mesh. This is the lowest end of the passivebasket end-effector.
Design and Integration of a Drone based Passive Manipulator for Capturing Flying Targets
Fig. 7: CAD model showing the idler pinion support assembly
The development of the end-effector is also influenced by the choice of the drone. Thedrone selected for testing is DJI M600. It was chosen as it fits within the size constraintsmentioned in Section 2 and also provides a flight time up to 30 minutes. The end-effectoris designed to be positioned at the side of the drone. This ensures safe detachment of theball, reducing the possibility of head-on crashes with the drone carrying the ball. Themanipulator is extended sideways via a rack and pinion mechanism. It is desired to haveminimum vibrations and play along horizontal and vertical planes. This is achieved byusing idler pinion gears with bearings shown in Fig. 7. The black acrylic plate holds themanipulator and is attached to the bottom of the drone frame. The manipulator extensionarm is a 15 mm ×
15 mm CF square tube of 1.2 m length. The idler pinions provides therequired tension while facilitating smooth actuation. In order to avoid vertical deflection,the rigidity of the arm is increased, which reduces the vibrations due to the deflectionexperienced by the arm.
5. Analysis
In this section, the proposed manipulator design is analysed. The major aspects examinedhere are the end effector dimensions, manipulator arm extension limits, location ofcamera, and the structural stability of the integrated system.
Considering the nature of the problem, two factors contribute to the successful grasping:grab volume and capture area. Grab volume is the effective volume available at theend-effector to capture the ball. Capture area is the effective area of the end-effectorthat engages with the ball to detach it from the target drone. The lower bound of grabvolume is defined by the size of the object to be grabbed. The volume constraint limitsthe maximum grab volume of the end-effector. The capture area is upper bounded by thesize constraints of the integrated system and lower bounded by the FoV constraints fromthe camera, which is described in the next section. Based on the shape of the end-effector,a truncated cone best represents the approximate volume of the end-effector. The topand front view of the passive basket marked with respective dimensions, are shown inFig. 8. The upper structure is approximated as a circle of diameter 0.51 m, as shown inFig. 8 (a). The capture area of the final end-effector is approximately a rectangle of sidesshown in Fig. 8 (b). The capture area is A passivecap = 0 . × .
175 = 89 . × − m .A ring size of 0.175 m is found feasible for the ball detector. The truncated coneas shown in Fig. 9a, is constructed with dimensions as given in Table I. Hence theapproximate grab volume is V passive g approx = 34.817 × − m The CAD model of the basketis shown in Fig. 9b. The precise grab volume is determined as 52 . × − m . A 33%increase from the approximate volume could be achieved in the final design by adjustingthe mesh shape to a accommodate a larger volume, during manufacturing. A majorfactor for the effectiveness of the design is the large capture area and grab volume, whichgreatly helps in handling small disturbances and oscillations of the ball due to wind ormaneuvers. esign and Integration of a Drone based Passive Manipulator for Capturing Flying Targets d d h h d h h d (a) (b) Fig. 9: (a) Approximate truncated cone for volume calculation (b) Precise grab volumedetermined from the CAD model of the final design
As described in Section 4, an eye-in-hand configuration is ideal for the proposed design.The location of the vision sensor is important for two reasons. The vertical placement ofthe camera should be such that the camera center and the ball center should coincide andthe ball should be within the basket. That is, if h c is the location of the camera belowthe basket top and r is the radius of the ball, then h c ≥ r . Larger object size wouldrequire larger h c which increases the size of the basket due to the FoV considerations.Considering these, the camera location is fixed at 0.15 m from the top of the basket.The schematic of how h c is measured from the top plane of the basket is shown in Fig.10(a). The second reason is that the location of the camera FoV should be free from anyobstructions. The scenario is shown in Fig. 10(b), where the ideal top and front viewof the basket is shown by the figures on the left and right, respectively. So, in order toensure a clear view, the minimum basket opening is d mincap = 2 h tan θ Design and Integration of a Drone based Passive Manipulator for Capturing Flying Targets hc (a) θh d cap (b) Fig. 10: (a) Camera positioning along the vertical plane (b) FoV considerations fordeciding the basket opening Table II : Total Impact workParameter Value (J) W Impact W Detach W Total h and θ are the planar depth and FoV angle, respectively, as shown in Fig. 10(b). One of the challenges mentioned in Section 2 is the impact on the manipulator andits ability to handle the detachment forces. The maximum drone velocity is v = 6 m/s.Impact strength is calculated in terms of work. Impact work is due to the kinetic energyof the ball. Detachment work is calculated as the detachment force times the detachmentdistance, which in this case is the diameter of the magnet. The net work is given in TableII. This net work is a representation of the total force. Effect of impact is determinedusing impact strength IS = W Total /A where, IS is the impact strength and A is the area opposing the impact. Fig. 11shows the forces involved in the detachment process. The cross-section of the detachingsinusoidal hull is 6 mm × W T is IS = 23 .
75 kJm − . Material selection was based on these impact calculationsand experiments performed on different end-effector prototypes. The final prototype ismanufactured using birch wood, which has an impact strength of 92.9 kJm − , whichensures adequate strength against the impact. Thus, the end-effector prototype is ableto perform effectively with minimum failure. a. Moment due to end-effector: The link joining the end effector and drone is designed asa linear actuated single DoF arm. Larger the extension, better is the safety factor. But,stability considerations impose a limit on the maximum extension of the manipulatorarm. The manipulator would act as a cantilever beam with the end-effector and the esign and Integration of a Drone based Passive Manipulator for Capturing Flying Targets × − m E 90 GPad 1.1 m δ max δ max = F × d × E × I where, δ max is the maximum deflection at the free end, E is the modulus of elasticityof the carbon fibre rod and I is the area moment of inertia due to the carbon fiber0 Design and Integration of a Drone based Passive Manipulator for Capturing Flying Targets
Fig. 13: Test setup for stationary and moving ball captureTable IV : Avionics and on board computers of test droneItem DetailsDrone M600 pro hex-rotorAuto pilot A3 proCompanion board NVIDIA Jetson TX2Auxiliary boards Arduino Mega, NanoVision module See3 130 HD cameraMiscellaneous Limit switchestube’s cross section. The values for these parameters and the maximum deflection isshown in Table III. The deflection calculation helps in orienting the camera with aslight tilt in the drone’s roll axis opposite to the sagging direction to compensate forthe error during flight.
6. Integration and Experimental Results
This section presents the details of the drones used and the associated test rigs for testingthe final manipulator prototype, and the results obtained.
Ball grabbing is tested by hanging a red ball of diameter 0.15 m under the DJI Mavic ProPlatinum (Fig. 13). The manipulator is tested for stationary drone as well as straightand curved paths of the drone carrying the ball. The control and vision modules forautonomous grabbing are the same as in. The hardware architecture and the control flow of the integrated system is shown inFig. 14. The avionics and the on-board computers are listed in Table IV. As shown in Fig.14, the ball is detected by the camera fixed at the center of the manipulator end-effector.This information is used by the the TX2 to compute control commands which is sent tothe M600 drone via A3 pro flight controller. The localisation is achieved via on-boardGPS. When the ball is detached, the limit switches are activated which, via the Arduinonano, sends the information that the ball is captured.The final prototype of the proposed manipulation mechanism was tested using DJIM600. The integrated system with the major components labelled, are shown in Fig. 15.
Flight tests are conducted in the test beds using the designed manipulators and drone setup at the airfield of Indian Institute of Science. The environment is windy which testedthe robustness of the designed system. The success rate of the manipulator for static ballis found to be 8/10 and maneuvering ball is found to be 7/10.A snapshot of a grabbing instance for static ball is shown in Fig. 16(a), where themanipulator moves towards the ball and pulls it to detach from the magnetic attachment. esign and Integration of a Drone based Passive Manipulator for Capturing Flying Targets A3 pro flight controller and GPSEnd-effector (eye-in-hand) See3 130 cameraTargetballLimitswitches Arduinonano Jetson TX2M600 hexacopter
Fig. 14: Hardware architecture of the integrated system
Camera Grab detectorManipulator end-effectorLinearactuatedmanipulatorOnboardcomputers
Fig. 15: M600 drone integrated with the manipulatorInstants of moving ball capture are shown in Fig. 16(b), where the ball is detached froma moving drone and the captured ball is collected in the mesh under the basket. Thedropping exercise using the release mechanism at the bottom of the basket end-effectoris shown in Fig. 16c. The drone approaches the box in which the ball is to be deposited,followed by actuating the servo to release the ball. The videos of the experimental resultsfrom which the above instants are captured, can be found in . Some interesting observations were made during the experiments. The visual feedback,grabbing algorithm, system dynamics, and the inherent delays in computation, werefound to have an impact on the success rate. This points to the unavoidable couplingbetween the manipulator design and the software used to perform the mission. Thereis a fine relationship between them which decides the success rate. The proper locationof the camera was also found to be an important parameter in successful grabbing,considering the field of view and approach direction. Effects of wind gust is minimal forthe proposed design, but any disturbance above 12 m/s brings in some vibrations. Sag ofthe manipulator arm was also observed over time, which was primarily due to the weightof wires at the end-effector side. Nevertheless, proper calibration of sensors resulted in https://youtu.be/1jdtIumUvdI Design and Integration of a Drone based Passive Manipulator for Capturing Flying Targets (a)(b)(c)
Fig. 16: Snapshots of grabbing of (a) static ball (b) moving ball (c) Snap shot of balldroppinggood success rate. With the present design, any object within 0.15 kg and within 0.2 mdiameter could be grabbed successfully.Few design and manufacturing aspects in the final design were compromised to fulfilthe requirements, which could be improved in the future variants. The shape of thetop portion was an intuitive design and gave good success rates, but a deeper analysiswould provide more effective shapes for similar end-effector footprint. The detachmentstructures and their locations could be analysed for better performance.
7. Conclusions
This work provided design, development and testing details of an aerial manipulationfor grasping of dynamic targets. The problem statement was inspired by Challenge1 of MBZIRC 2020. The major complexities involved in the design and developmentare discussed in detail. The conceptual design for the task was presented, describingthe reasons for design and the material choice. The experimental results and relevantobservations are also reported in this paper. The obvious and unavoidable couplingbetween the hardware design and software modules are pointed out. The repeatabilityand success rates of the final configurations are reported and possible developments onthe final design are also presented. The manipulator has been designed for a ball grabbingin this work but can be modified for many other applications.
Acknowledgements
We acknowledge members of GCDSL for their valuable suggestions in developmentof the manipulator. The contributions of Integrative Multiscale Engineering Materialsand Systems (iMEMS) lab and Advanced Materials and Processing Laboratory esign and Integration of a Drone based Passive Manipulator for Capturing Flying Targets
Author Contributions
Author A designed the manipulator. Authors AB prototyped the manipulator and wrotethe paper. Author C was the pilot for the tests and dealt with the hardware integration.Authors DEF carried out the software integration to automate the process and helpedwith the tests. Author G conceived the design idea and wrote the paper. Author F leadthe project, reviewed work progress and wrote the paper.
Financial Support
This work is partially supported by Robert Bosch Center for Cyber Physical Systems,(IISc) and Khalifa University, Abu Dhabi, UAE.
Competing Interest Declaration
Competing interests: The author(s) declare none.
References
1. S. Waharte, and N. Trigoni, “Supporting search and rescue operations with UAVs,” In
InternationalConference on Emerging Security Technologies , pp. 142-147, (2010, September).2. Amazon,“Amazon prime air”, , accessed on 3 February 2021.3. Zipline, “Zipline Vital, On-Demand Delivery for the World”, , accessed on 12December 2020.4. DJI, “DJI - Official Website”, , accessed on 5 January 2021.5. Skydio,”Skydio”, , accessed on February 2021.6. F. Ruggiero, V. Lippiello, and A. Ollero, “Aerial manipulation: A literature review.”
IEEE Roboticsand Automation Letters , 3(3), pp. 1957-1964, (2018).7. X. Meng, Y. He, and J. Han, “Survey on aerial manipulator: System, modeling, and control,”
Robotica ,38(7), pp. 1288-1317, (2020).8. M. Orsag, C. Korpela, S. Bogdan, and P. Oh, “Valve turning using a dual-arm aerial manipulator,”In , pp. 836-841, (2014).9. R. Boudreau, S. Nokleby, and M. Gallant, “Wrench Capabilities of a Kinematically Redundant PlanarParallel Manipulator,”
Robotica , pp. 1–16, (2021).10. Jeong, Donghwa, and Kiju Lee. ”Design and analysis of an origami-based three-finger manipulator.”Robotica 36, no. 2 (2018): 261-274.11. K. Kondak, F. Huber, M. Schwarzbach, M. Laiacker, D. Sommer, M. Bejar, and A. Ollero, “Aerialmanipulation robot composed of an autonomous helicopter and a 7 degrees of freedom industrialmanipulator,” In , pp. 2107-2112, (2014).12. M. Fumagalli, R. Naldi, A. Macchelli, F. Forte, A. QL Keemink, S. Stramigioli, R. Carloni, and L.Marconi, “Developing an aerial manipulator prototype: Physical interaction with the environment,”
IEEE robotics and automation magazine , 21(3), pp. 41-50, (2014).13. C. D. Bellicoso, L. R. Buonocore, V. Lippiello, and B. Siciliano, “Design, modeling and control of a5-DoF light-weight robot arm for aerial manipulation,” In , pp. 853-858, (2015).14. Y. Wu, J. Song, J. Sun, F. Zhu, and H. Chen, “Aerial Grasping Based on VR Perception and HapticControl,” In
IEEE International Conference on Real-time Computing and Robotics (RCAR) , pp. 556-562, (2018).15. G. Loianno, and V. Kumar, “Cooperative transportation using small quadrotors using monocularvision and inertial sensing,” In
IEEE Robotics and Automation Letters , 3(2), pp. 680-687, (2017).16. A. Tagliabue, M. Kamel, S. Verling, R. Siegwart, and J. Nieto, “Collaborative transportation usingmavs via passive force control,” In
IEEE international conference on robotics and automation (ICRA) ,pp. 5766-5773, (2017).17. G. Gioioso, A. Franchi, G. Salvietti, S. Scheggi, and D. Prattichizzo, “The flying hand: A formationof UAVs for cooperative aerial tele-manipulation,” In
IEEE International conference on robotics andautomation (ICRA) , pp. 4335-4341, (2014). Design and Integration of a Drone based Passive Manipulator for Capturing Flying Targets
18. S. Hamaza, I. Georgilas, and T. Richardson, “An adaptive-compliance manipulator for contact-basedaerial applications,” In
IEEE/ASME International Conference on Advanced Intelligent Mechatronics(AIM) , pp. 730-735, (2018).19. Ikeda, Takahiro, Shogo Yasui, Motoharu Fujihara, Kenichi Ohara, Satoshi Ashizawa, Akihiko Ichikawa,Akihisa Okino, Takeo Oomichi, and Toshio Fukuda. “Wall contact by octo-rotor UAV with one DoFmanipulator for bridge inspection.” In 2017 IEEE/RSJ International Conference on Intelligent Robotsand Systems (IROS), pp. 5122-5127. IEEE, 2017.20. S. Kim, H. Seo, S. Choi, and H. J. Kim, “Vision-guided aerial manipulation using a multirotor witha robotic arm,”
IEEE/ASME Transactions On Mechatronics , 21(4), pp. 1912-1923, (2016).21. S. Kim, H. Seo, J. Shin, and H. J. Kim, “Cooperative aerial manipulation using multirotors withmulti-dof robotic arms,”
IEEE/ASME Transactions on Mechatronics , 23(2), pp. 702-713, (2018).22. H. Seo, S. Kim, and H. J. Kim, “Aerial grasping of cylindrical object using visual servoing based onstochastic model predictive control,” In
IEEE international conference on robotics and automation(ICRA) , pp. 6362-6368, (2017).23. C. C. Kessens, J. Thomas, J. P. Desai, and V. Kumar, “Versatile aerial grasping using self-sealingsuction,” In
IEEE international conference on robotics and automation (ICRA) , pp. 3249-3254, (2016).24. C. C. Kessens, M. Horowitz, C. Liu, J. Dotterweich, M.Yim, and H. L. Edge, “Toward lateral aerialgrasping and manipulation using scalable suction,” In
International Conference on Robotics andAutomation (ICRA) , pp. 4181-4186, (2019).25. C. Liu, A. Bera, T. Tsabedze, D. Edgar, and M. Yim, “Spiral Zipper Manipulator for Aerial Graspingand Manipulation,” In
IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) ,pp. 3179-3184. (2019).26. G. Zhang, Y. He, B. Dai, F. Gu, L. Yang, J. Han, G. Liu, and J. Qi, “Grasp a moving target fromthe air: System and control of an aerial manipulator,” In
IEEE International Conference on Roboticsand Automation (ICRA) , pp. 1681-1687, (2018).27. MBZIRC 2020, “Mohammad Bin Zayed International Robotics Challenge 2020”, , accessed on 12 December 2020.28. B. V. Vidyadhara, L. A. Tony, M. S. Gadde, S. Jana, V. P. Varun, A. Bhise, S. Sundaram, and D.Ghose, “Design Iterations for Passive Aerial Manipulator,” arXiv preprint arXiv:2102.08306 (2021).29. DJI, “DJI M600”, , accessed on 10 November 2020.30. Russian Birch Lumber, “Russian Birch Lumber - About Birch”, , accessed on 5 November 2020.31. Performance Composites Ltd, “Mechanical Properties of Carbon Fibre Composite Mate-rials, Fibre / Epoxy resin (120 ° C Cure)”,