IROS2019国际学术会议论文集 2331

Abstract A flexible blade that is easy to fabricate and repair is presented It has a low impact load when colliding with obstacles and can generate enough thrust for drone flight The flexible blade has a camber structure that mimics the feature of dragonflies to increase wing stiffness It makes the blade s deformation decline when it receives a lift force and drag force A commercially available drone that conformed to standards was used to design fabricate and test the flexible blade The flexible blade consists of a pivot column made by 3D printer and a polyethylene film The design variables were defined as values with thrust and efficiency that were determined to be similar to those of the original blade through experiments Experiments for safety verification were conducted by comparing the impact and sharpness of the original and the flexible blades Experimental results show that the flexible blade s impact force is approximately a tenth of the original blade s impact force and only the original blade damaged a 100 m film obstacle In general terms the suggested flexible blade is safer and more economical than a plastic blade in addition to having performance close to that of the original blade during flight I INTRODUCTION Safety has become an important issue in the robotics field with increasing demand for robotics in daily life during in recent years In particular a drone has many interactions with people for variety purposes such as playing photographing research and reconnaissance Although drones are very useful and provide many conveniences they have numerous risks such as a person s injury and damage to property from a blade or from pieces of a broken drone 1 Many studies have been investigated to avoid hazards from drone accidents For example the propeller protection which prevents direct contact between a blade and an obstacle protects a blade from an impact 2 3 A collision avoidance system that detects an obstacle and creates a detour has also been developed for unmanned aerial vehicles UAVs following many approaches 4 6 Also an electromechanical stopper system was studied This system forces a propeller to stop by electronic braking when the propeller collides with any obstacles 7 However these methods do not provide solutions for the immediate cause of the collision The propeller protection for safety although it protects a drone from obstacles This work is supported by KITECH R fax 82 31 8040 6380 e mail yanggh kitech re kr approaching from the side imposes extra weight on a drone Sometimes it cannot prevent injury when pilots grab the lower part or upper part of the drone A collision avoidance system with a camera can check on an immovable or slow moving object However the system cannot avoid obstacles that are not in the angle of view or that move fast and it makes the drone have a narrow flight area In the past few years most studies of robots that interact with people had focused on control systems 8 9 or compliance mechanisms 10 11 to improve the safety of robot Although the suggested methods improve the safety users are still exposed to dangerous situations because of the heavyweight and hardness of a rigid robot Recently the notion of a soft robot has been suggested to solve the problems that rigid robots have 12 15 The soft robot is composed of soft materials such as silicon polyester or paper Because soft materials make the robots light weight and flexible a robot with soft materials has high safety by means of a passive compliance system and low impact force Flexible blades have been studied to transform the length of a wing by rolling into a pivot column and to decrease the wing weight The most representative one is the flexible blade developed in the University of Texas at Austin 16 17 The blade entirely consists of a fabric membrane and carbon material that is heavier than the membrane is combined on the tip of the blade This helps the blade to stay in shape when the wing rotates during flight In this work however an attempt was made to solve the immediate cause of a rigid blade and to improve the safety of the rotor based UAVs The flexible blade has less sharpness and low impact force when a collision occurs because it possesses both passive compliant structure and low weight because of the soft material In addition the proposed flexible blade is easier to fabricate repair and customize than a rigid blade Because the rigid blade is difficult to fabricate with a 3D printer and repair users have to purchase an additional blade or make a blade using a carbon flat sheet which is more expensive and dangerous than a plastic one However the flexible blade reduces the extra cost and extra time because it consists of a pivot column constructed by a low quality 3D printer and affordable film such as polyethylene PET film II CONCEPT OF FLEXIBLE BLADE The flexible blade should have sufficient rigidity in the direction of the force and should not be stretchable Because the amount of thrust and torque is relative to the area when a blade rotates consistently the blade has to maintain the area that contacts air to generate sufficient thrust However the flexible material easily bends when the force is exerted to the surface because it is thin and has low rigidity Therefore in Design and Experimental Study of Dragonfly Inspired Flexible Blade to Improve Safety of Drones JaeHyung Jang Kyunghwan Cho and Gi Hun Yang IEEE Robotics and Automation Letters RAL paper presented at the 2019 IEEE RSJ International Conference on Intelligent Robots and Systems IROS Macau China November 4 8 2019 Copyright 2019 IEEE order to maintain the flexible blade s shape for rotor based air vehicles the rigidity in the direction of the force should be increased In this study the rigidity of a flexible blade was improved by inspiring the characteristics of insects wing such as those of dragonflies flies and beetles In particular dragonflies wings although they have a very thin membrane 5 10 m and main vein 40 90 m can endure relatively large force 18 19 These results suggest the corrugation wing structure s possibility to bear large aerodynamic and inertial force The flexible blade imitates the dragonflies corrugated wing shape to reinforce the rigidity of the wing While the vein of dragonflies increase the wing rigidity of upper and lower areas the flexible blade needs to increase blade stiffness of the direction receiving thrust and torque without a rigid part Fig 1 shows the amount of the flexible material s stiffness through the bending generated by self load when the material has a camber and when it has a flat structure Comparing the two structures shows that the material of flat structure in Fig 1 a bends toward the gravity direction while Fig 1 b illustrates that the flexible material stays in straight shape with increasing its stiffness when it is warped against the gravity direction The stiffness of the material shown in Fig 1 c also increases more than that of the flat structure as the curvature of the material is generated toward the gravity direction but it is not as highly rigid as the camber structure is against the gravity direction 20 In addition the wing with camber structure has higher efficiency than those with flat structure 21 Thus this research proposes a flexible blade having a camber structure to improve safety by applying it to a rotor based air vehicle Figure 1 Comparative experiment between a flat structure and bending structure 20 III DESIGN AND FABRICATION In this study Rodeo 150 Walkera Inc shown in Fig 2 a 22 was used as a standard drone to design the flexible blade because the blade design parameters need a target value Figure 2 Rodeo 150 drone a with an original blade and b with a flexible blade and Phantom 4 a with an original blade and b with a flexible blade In addition PHANTOM 4 DJI Inc shown in Fig 2 c was designated as the application drone to confirm that the defined design parameters rate can be applied to a drone weighting more than 1kg 23 The specifications of the drones are indicated in Table 1 and drones with fabricated flexible blades are illustrated in Figs 2 b and d Each blade should generate at least 55 g thrust to hover because the Rodeo drone s weight was 220g in the first experiment to define the design parameters Therefore the flexible blade had to endure the force when the blade generated over 55g of thrust by maintaining the shape The specifications of the drones are indicated in Table 1 and drones with fabricated flexible blades are illustrated in Figs 2 b and d Each blade should generate at least 55 g thrust to hover because the Rodeo drone s weight was 220g in the first experiment to establish the design parameters Therefore the flexible blade had to endure the force when the blade generated 55g of thrust by maintaining the shape TABLE I SPECIFICATION OF RODEO 150 DRONE Component Specification Rodeo 150 Specification Phantom 4 Rotor diameter 96 mm 300 mm Weight battery included 220 g 1380 g Dimension 137 148 77 mm 550 550 400 mm Battery 7 4 V 850 mAh 15 2 V 5350 mAh Motor WK WS 17 002 Original 2312S A Designing pivot column and wing The design parameters to develop the blade are indicated in Table 2 and Fig 3 The flexible blade has the same aspect ratio and width as the original blade of Rodeo 150 Each of the design parameters was determined by the affecting angle of attack efficiency and rigidity To be specific Lf and Lb determine the bending position of a blade which influences the wing s efficiency d transforms the angle of attack and leading edge angle and is a design parameter affecting the angle of attack and stiffness of a blade Fig 3 a provides an isometric projection of the blade combined with a pivot column for these main factors Fig 3 b which indicates the cross section between a pivot column and a blade shows both main factors The blade was designed to become shorter toward the tip as shown in Fig 3 c The flexible blade has the lowest stiffness at the tip 20 because the tip of the wing moves faster than the root the blade s tip receives the largest force The camber was made in 6 10 37 5 position because the efficiency of the blade is optimal when the structure s camber is located in 30 45 24 Film thickness t d and were determined experimentally as described in Section 4 Fig 4 shows the designed pivot column and wing The pivot column s attaching part for the wing sticks to the front and back areas of the wing because attaching the middle of the root area to the pivot column makes the efficiency of the wing worse by reducing the thrust generating area In addition the torque from a thrust occurs when the wing is just attached in the front and back areas with a pivot column To stop the rotation of wing by torque the pivot column is designed to be higher than the root of wing as a stopper Fig 4 shows the designed pivot column and wing The pivot column s attaching part for the wing sticks to the front and back areas of the wing because attaching the middle of the root area to the pivot column makes the efficiency of the wing worse by reducing the thrust generating area In a b c a b c d addition the torque from a thrust occurs when the wing is just attached in the front and back areas with a pivot column To stop a rotation by torque the pivot column has a greater length than the wing as a stopper Figure 3 a Equal angle view of pivot column and wing b the cross sectional view of the wing with respect to the vertical direction and c the top view of the wing for design parameters Figure 4 Designed pivot column and wing in equal angle view TABLE II THE BLADES DESIGN PARAMETER SPECIFICATIONS Design parameter Specification Rodeo 150 Specification Phantom 4 15 15 120 120 6 mm 10 mm 10 mm 20 mm L 16 mm 30 mm 10 mm 20 mm 43 mm 86 mm t 100 m 400 m AR Aspect ratio 3 3 B Fabrication One of the study s purposes was to make a wing with a low quality 3D printer for a self produced drone Only the body part of a self produced drone has been manufactured with a 3D printer previously because it is difficult to make the blade part with a 3D printer A carbon plate is sometimes used to make a blade but it is more expensive and dangerous than a plastic blade Therefore a low quality 3D printer Zortrax M200 resolution 100 m and PET film were used to verify that the flexible blade can be fabricated with a low quality 3D printer that is used as a diffusion device Fig 5 shows a fabricated wing and an original wing The fabricated wing consists of a pivot column for mounting a blade and PET film blade The fabricated wing has a similar length to the original blade Figure 5 Comparision between fabricated wing and original wing IV EXPERIMENT TO MEASURE THRUST AND EFFICIENT A Test setup A rotor actuator and a force measurement sensor were used to measure the thrust and torque Fig 6 shows the force measurement setup The experiment setup consist of a brushless DC BLDC motor from Maxon 13 brushless 50 W Hall sensors a Nano 17 Force Torque sensor resolution force 0 0125 N torque 0 0625 Nmm an ESCON 36 3 EC controller Maxon a NI DAQ USB 6343 and a gearbox with a 1 4 gear ratio Figure 6 A force measurement equipment Fig 7 shows the block diagram of data acquisition The angular velocity and samples per second were entered in the user interface as the input data with LabVIEWTM software The data acquisition device sent an angular velocity of motor data to the ESCON controller on the basis of input data The ESCON controlled the angular velocity using a closed loop control whose input was position data sent by the motor s encoder sensor The velocity force and torque data which were measured by the F T sensor with 10 000 data per second were recorded as text file afterward the data were used for calculating the thrust torque and efficiency of the blade by averaging 20 000 data Leading edge b x z Attached area with pivot column c x y x z y a Force Torque Wing Pivot column Stopper 96mm Flexible blade Pivot column Blade Pivot column Gear box Maxon motor F T sensor Figure 7 The block diagram of data acquisition to measure force and torque of a blade B Experiment to measure thrust and efficient experiment The thrust experiment was conducted through four steps The first step was to determine the film s thickness The blade s stiffness is the most important factor for a flexible blade to maintain the blade s shape The film s thickness and affect the stiffness However because affects both the stiffness and angle of attack the first step proceeded to determine the film s thickness which only influences the stiffness In the second step and the third step a value was defined The stiffness and the angle of attack are inversely related to and the angle of attack less than 45 is proportional to the lift force coefficient and drag force coefficient as shown in the Equations 1 and 2 which are formulations of both a lift force coefficient in Equation 1 and a drag force coefficient in Equation 2 in a low Reynolds number condition 25 In the equations and are the lift force coefficient and drag force coefficient and is angle of attack CL 0 225 1 58sin 2 13 7 2 1 CD 1 92 1 55cos 2 04 9 82 2 Figure 8 Wing cross section with respect to vertical direction and variables for aerodynamics is actual angle of attack Vb is a blade speed Vf is a vehicle speed is rotation angle of attack and is direction of vehicle s movement The angle of attack for a blade is defined by the rotation speed of the blade and the vehicle s speed Fig 8 shows the wing cross section to define the relationship between the angle of attack and the velocity of the blade In the figure Vb is blade speed and Vf is vehicle speed Therefore which is the actual angle of attack becomes a difference between angle of attack for rotation of the blade and direction of the vehicle s movement After defining the variables in a quasi steady state the lift force and drag force are calculated by Equations 3 and 4 through the measured lift and drag coefficients FL 1 2 CL AV 2 3 FD 1 2 CD AV 2 4 The thrust is expressed by Equation 5 by the lift and drag force where is the blade s pitch angle 26 T FLcos FDsin In the study is 0 because of measuring the force of the fixed motor and blade Thrust is derived as Equation 6 by substituting the 0 for Equation 5 T FL 6 Therefore is tuned for finding a similar thrust to that of the original blade while maintaining the blade s shape In the final step the that makes the blade have a thrust and efficiency close to that of the original blade which was determined Fig 9 shows that the measured original blade s thrust with the BLDC motor used in the experiment had more than a 55g thrust between 10 000 and 11 000 RPM Figure 9 Thrust angular velocity relationship of original blade 1 Experiment to define a thickness To determine the film thickness the blades were fabricated with 0 d and 90 the experiments were conducted for 25 50 and 100 m film thicknesses All experiments were conducted with an angular velocity interval of 1000 RPM Fig 10 shows the graph for the measured thrust angular velocity relationship by first experimenting from 0 to 8000 RPM The films with 25 and 50 m thicknesses could not maintain their shape at 2000 and 6000 RPM respectively while the 100 m thickness film retained the shape up to the motor torque limit of 8000 RPM Three types of flexible Angular velocity Sample per second User Interface using LabVIEW DAQ ESCON PWM Nano17 Actuator Closed loop controlPosition Angular velocity DAQ Force torque txt Aerodynamic analysis Force torque Angular velocity Averaging Input Measurement Data acquisition the shape of the blade designed with a of 150 could not be maintained at 8000 RPM Fig 11 b shows the result of the experiment between 7000 and 10 000 RPM by adding 120 A flexible blade has the most similar thrust to that of the original blade at a of 130 but the value was determined to be 120 which has higher stiffness and generates more thrust than 55 g as the thrust decreases when the ang