An Experimental Characterization of a 1-DOF Anthropomorphic Arm for Humanoid Robots

Size: px

Start display at page:

Download "An Experimental Characterization of a 1-DOF Anthropomorphic Arm for Humanoid Robots"

Maximillian Blake
5 years ago
Views:

1 Proceedings of the th WSEAS International Conference on SYSTEMS An Experimental Characterization of a -DOF Anthropomorphic Arm for Humanoid Robots HAO GU, MARCO CECCARELLI and GIUSEPPE CARBONE LARM: Laboratory of Robotics and Mechanics DIMSAT, University of Cassino Via G. Di Biasio, 4, Cassino, 4 ITALY {hao.gu, ceccarelli, carbone}@unicas.it Abstract: - In this paper, we have reported results of the experimental tests for validating and characterizing a new design and its operation for a humanoid robot arm with one actuator only. The core of the arm system consists of clutches that are activated or deactivated according to a programmed manipulation path-planning. Experimental tests are discussed with measured results to characterize the functionality of the proposed clutch system for a robotic arm by looking at stationary and impulsive response of the clutch motion transmission. Key-Words: - Humanoid Robots, Anthropomorphic Arm, Clutch Systems, Experimental Characterization Introduction A human arm is a research topic of still great interest and it is studied by robotic or biomechanical specialists, who effort to understand and model the human arm both for analysis and design purposes [][][][4][5]. Nowadays, for humanoid or service robots, arms are still critical parts, and they are expected to be more anthropomorphic, lower energy consumption and safer. For these purposes, many researchers have developed designs for anthropomorphic arms like examples in [6][7][8][9][][]. Famous examples are the robotic arms in ASIMO[] and WABIANR[]. They act quite smoothly, flexibly, and human likely, but they are also very expensive and complex systems for no expert users. At LARM (Laboratory of Robotics and Mechatronics, University of Cassino), a research line has been devoted to design low-cost and easy-operation solutions for humanoid robots. Recently, a new robotic arm has been proposed in [4] for a humanoid robot called CALUMA (CAssino Low-cost humanoid robot) [4][6]. In order to check the feasibility of the design from practical point of view before building a prototype, an experimental characterization of the mechanical design have been carried out through experimental tests of the clutched solution, which is the core of the design. A New Robotic Arm Since the number of actuators affects significantly the cost and the operation complexity of a robotic arm, the guide line for arm design at LARM is to include as few actuators as possible. As shown in Fig. and Fig., the proposed new robotic arm has only one motor, but it can move in a D space [4]. The three rotations of the arm motion have been defined as R, R and R respectively. θ toθ are the rotating angles. L, L and L are the length of the clavicle link, upper arm and forearm, respectively. L is the height of the shoulder as assembled in a trunk body. A parallelogram-based design makes rotations R and R have the same rotation axis. This feature makes it possible to get the motions from one motion source only, and it can let the arm structure light and stiff as attempted in [7]. θ θ θ Fig. A kinematic scheme of the proposed new LARM arm. ISSN: ISBN:

2 Proceedings of the th WSEAS International Conference on SYSTEMS complete the low-cost easy-operation features for the overall arm system. Fig. The CAD model of the new LARM arm in ADAMS: a model with all reference frames; a zoomed view of the shoulder. Fig. and Fig. shows a mechanical design for a clutch system in the shoulder mechanism, which is the key part for generating the three rotations R, R and R by one motor. It is a gearing system consisting of spur gears, bevel gears, worms and worm wheels. In the schemes in Fig., symbols Gia/b (i=-6) represent gears while Wia and Wib (i=,) represent the worm and the worm wheel, respectively. Clutches that are named as Ca, Cb, C and C are assembled on the rotating axis for each DOF of the robotic arm. By changing the states of the clutches, the gearing system can be activated in one or more transmission lines. Table lists the operation modes for the arm, where means that a clutch is disconnected and a clutch is connected. Thus, these operations can be easily controlled by switch actions. For example, a PLC (Programmable Logical Controller) can be a suitable low-cost controller for this solution to Wb Wb G5a G5b G6a G6b Fig. A scheme for the mechanical design of the shoulder mechanism in Fig. : front view; lateral view. Table The operation modes for the arm in Fig. and Fig.. Operation mode Active rotations Ca Cb C C OP R OP R OP R OP4 R&R OP5 R&R OP6 R&R OP7 R&R&R ISSN: ISBN:

Proceedings of the th WSEAS International Conference on SYSTEMS With the proposed seven operation modes named OP to OP7, the arm can move like having three DOFs.

4, in a frame with three axes of θ, θ andθ, the path possibility for operation modes can be expressed as vectors from an initial configuration withθ, θ, θ zero

The operation path and the trajectory of the arm tip are shown in Fig.

θ (deg) 4-4 4-4 θ (deg) OP7 (R&R&R) -4 4 θ (deg) Fig. 4 The path possibility of operation modes in Table.

Clutches, worms and worm wheels are the critical parts, since clutches make the only motor to be shared by three rotations, while the worm and worm wheel

Thus, novel merits of the new robotic LARM arm in terms of easy operation, low energy consumption and avoiding overheat of the motor are based on the

Before building a prototype of the arm, operating it and installing it on a humanoid robot, we have thought convenient to get more practical information about

3 Proceedings of the th WSEAS International Conference on SYSTEMS With the proposed seven operation modes named OP to OP7, the arm can move like having three DOFs. As shown in Fig. 4, in a frame with three axes of θ, θ andθ, the path possibility for operation modes can be expressed as vectors from an initial configuration withθ, θ, θ zero degree. In a previous paper [4], we have checked three basic operation modes by simulating them in ADAMS environment. The operation path and the trajectory of the arm tip are shown in Fig. 5, in which the arm rotates 45 degrees on R, R and R sequentially while the positive direction of the angle is assumed counterclockwise. θ (deg) θ (deg) OP7 (R&R&R) -4 4 θ (deg) Fig. 4 The path possibility of operation modes in Table. A Test Bed The feasibility and efficiency of the proposed arm design strongly depend on the operation characteristics of the clutch transmissions. Clutches, worms and worm wheels are the critical parts, since clutches make the only motor to be shared by three rotations, while the worm and worm wheel combination may eliminate additional locking mechanisms. Thus, novel merits of the new robotic LARM arm in terms of easy operation, low energy consumption and avoiding overheat of the motor are based on the performances of clutches and worms as efficient and effective as described theoretically. Before building a prototype of the arm, operating it and installing it on a humanoid robot, we have thought convenient to get more practical information about the functionality of the clutch system. Therefore, it has been thought convenient to have a test bed for understanding the assembly and the operation of commercial clutches and also for checking the self-locking feature of worms. Fig. 6 shows the mechanical design and the prototype of the test bed, which mainly consists of a motor, a clutch and a pair of worm and worm wheel. This is the basic transmission chain in the arm s mechanical design in Fig.. θ (deg) θ (deg) θ (deg) End R Z Y X R R Start Fig. 5 An operation mode as simulated in [4]: the operation path; the trajectory of the arm tip. Fig. 6 A test bed for a clutch transmission: a scheme; a prototype. ISSN: ISBN:

4 Proceedings of the th WSEAS International Conference on SYSTEMS Table lists main components that are used in the test bed for experimental tests. Table Components of the test bed in Fig. 6. Components Attributes motor 4VDC, 5.9:Ratio,.Nm clutch 4VDC,.7Nm worm Module, deg Pressure angle worm wheel Module, 4 teeth PLC SIEMENS LOGO!/4RC accelerometer ADXL5AQC single axis A Siemens PLC is used for operating the clutch and the motor. Fig. 7 gives the circuit diagram and the PLC wiring, in which Si (i=,) are switches, Qi (i=) are the outputs of the PLC, which can be considered as programmable relays. As Fig. 7 shows, the clutch is switched on and off by the switch S, while the rotation direction of the motor is changed by switch state of S. value that can characterize both kinematic and dynamic performance of the basic clutch unit of the proposed arm design. The aim of the experiments on the test bed is to check the performance of the clutch (response to the switch action and transmitting torque) and the self-locking feature of the worm and worm wheel combination. 4 Experimental Results For the experiments, the test bed is operated with the motor working at a constant speed and changing the state of the clutch instantaneously. Fig. 8 A scheme for the paylod motion. + - S S motor M S C clutch clutch operation signal Fig. 7 A scheme for the operation of the test bed via PLC: the circuit diagram; the wiring for SIEMENS LOGO! PLC. An accelerometer is installed on a payload, and a NI (National Instruments) data acquisition (DAQ) device is used for getting the experimental results through a proper virtual instrument. The of the payload is measured in operation modes as a As show in Fig. 6, the motion of the payload is designed to move vertically through a suitable pulley. A clear scheme is given in Fig. 8, in which T is the output torque, r is the radius of the pulley, m is the payload mass, and a is the measured. By using this payload motion, a single axis accelerometer can be successfully used in the experiments, since by analyzing properly the values all the interested information can be deduced. Considering different operation modes for the payload and the motor speed, four groups of experiments have been carried out. Each test mode includes two experiments with different moving directions. As listed in Table, Eia (i=-4) represents the payload moving up, while Eib (i=-4) indicates payload moving down. Table Operation modes of experimental tests with test bed in Fig. 6. Motor Speed N. test Payload Duration voltage direction Ea up s E 5V Eb down.6s.4kg Ea up.s E 4V Eb down 4s Ea up 5.s E 5V Eb down 4.5s kg E4a up.s E4 4V E4b down 5.4s ISSN: ISBN:

5 Proceedings of the th WSEAS International Conference on SYSTEMS The acquisition frequency is 5 Hz and the results are reported in Fig. 9 to Fig.. Besides the, the operation signal for the clutch has been also captured to offer a clear view for understanding the state of the clutch. As shown in Fig. 9 to Fig., the curve has the value either or to indicate inactive or active stated of the clutch. The duration column in Table is the time for keeping the clutch connected, which has been represented as the width of the square wave in Fig. 9 to Fig.. As shown in Fig. 9 to Fig., when the clutch is activated instantaneously, there is a shock of the. When the clutch is released, there is a shock again. The first shock is due to the fact that the clutch transmits the motion from the motor immediately, while the second shock indicates the effeteness of self-locking of the worm and worm wheel combination. Accordingly, in Eia (i=-4), because of the upward motion of payload, the first peak for the clutch activation is positive while that for clutch release is negative. In Eib (i=-4), because of moving down, the situation is opposite. (m/s ) (m/s ) Fig. 9 Result for experimental test E in Table : Ea; Eb. (m/s ) (m/s ) Fig. Result for experimental test E in Table : Ea; Eb. (m/s ) (m/s ) Fig. Result for experimental test E in Table : Ea; Eb. ISSN: ISBN:

6 Proceedings of the th WSEAS International Conference on SYSTEMS (m/s ) (m/s ) Fig. Result for experimental test E4 in Table : E4a; E4b. The maximum s for E and E with a.4kg payload are.8m/s and.5 m/s respectively, which is larger than.7 m/s and m/s for E and E4 with a kg payload. The output torque of the motor is fixed with the same input voltage, so that once the payload mass is larger, its becomes smaller. Additionally, E and E4 show more vibrations than E and E when the clutch works in a stationary state. It is due to the larger friction between worm and worm wheel with kg payload. As regarding the effect of the motor speed, it is reasonable to get the results that the peak of E (.5m/s ) or E4 (m/s ) with 4V motor voltage is larger than E (.8m/s ) or E (.7m/s ) with 5V. A zoomed view for the clutch activation of test Ea is shown in Fig.. In order to better appreciate the performance of the payload motion, a velocity curve is added by integrating the values. There is a delay between the change of the clutch operation action and the change of the payload motion. This is the activation time of the clutch, which can be an important parameter for operating the new LARM arm. If the -DOF arm can change its active rotation fast enough, the arm will move as smoothly as having more actuators. (m/s ) velocity Fig. A zoomed view when clutch is activated in the test Ea in Fig. 9. The activation time of clutch in Fig. has been evaluated as 6ms. Of course, because of the working feature of electromagnetic clutches, the activation time can be affected by the input voltage and the assembling. However, to have the reaction of approximate milliseconds can be suitable for the arm s applying, when it is compared with the duration for the arm to complete a planned motion task. (m/s ) velocity Fig. 4 A zoomed view when the clutch is releaed in the test Ea in Fig. 9. Fig. 4 shows the zoomed view for the clutch release in the test Ea. The motion is stopped by the self-locking of the worm and worm wheel combination. Nevertheless, some time is needed for the clutch release and it is evaluated in 4ms, which is longer than the activation time in Fig.. This delay is due to the inertia of the rotating worm and clutch. Similarly, the delay time at clutch has been evaluated as ms, 4ms, 56ms, ms, ms, ms, 4ms while the delay time at clutch deactivation as 6ms, 8ms, 4ms, 7ms, 58ms, 48ms, 44ms for the ISSN: ISBN:

7 Proceedings of the th WSEAS International Conference on SYSTEMS tests in Fig. 9, Fig., Fig., Fig., Fig., Fig., Fig.. Referring to the scheme in Fig. 8, the output torque of the test bed can be calculated by using the expression T=m(a+g)r () in which, g is the of the gravity; a is the due to the motor action; m is payload mass; and r is the radius of the pulley. Thus, torque results of test E4a in Fig. have been computed as shown in Fig. 5. The upward motion with a kg payload and high speed (4V motor voltage) requires the largest output torque of the test bed in all the tests listed in Table. Fig. 5 shows the outputs with a.nm peak at beginning, a.5nm peak at end and a.9nm average torque during stationary in the test E4a. torque(nm) torque Fig. 5 The output torque of the test bed in the test E4a. From test results, as shown in the plots of Fig. 9 to Fig. 5, which have been useful to check the feasibility of the clutch system by using commercial components with the following characteristics: quick activation of the output motion through activation of the clutch the release delay in deactivation phase of the clutch does not affect significantly the efficiency of the output motion the stationary performance shows constant behavior with limited energy loss and vibration noise as much as longer is the stationary operation the damping of the shocks at the activation and deactivation show short time duration the shock at deactivation of the clutch is consequent also of the self-locking behavior of the worm and worm wheel combination, which gives a firm axle configuration likewise a brake. the transmission ratio from the worm and worm wheel to the output axis contributes significantly to use small sized motor and clutches. 5 Conclusion A new robotic arm has been proposed with one actuator only but with D motion ability through a clutch system. Since clutches and worm combinations are the core parts for achieving most of the merits of the arm, a test bed has been built for checking the practical performance and characterizing the arm functionality. Experimental results show that the selected commercial clutch has a quick response and an enough transmitting ability while the worm and worm wheel combination has a reliable self-locking feature. Vibration noise has been detected with values that do not affect considerably the manipulation capability of the arm. Acknowledgements The first author would like to acknowledge China Scholarship Council (CSC) for supporting his PhD study and research at LARM in the University of Cassino, Italy, for the years 8. References: [] J. Lenarcic and A. Umek, Simple model of human arm reachable workspace, IEEE Transactions on System, Man, and Cybernetics, Vol.4, No.8, 994, pp.946. [] T. B. Moeslund, C. B. Madsen and E. Granum, Modeling the D pose of a human arm and the shoulder complex utilizing only two parameters, Integrated Computer-Aided Engineering Vol., No., 5, pp [] M. E. Rosheim, Robot evolution - The development of anthrobotics, Wiley, New York, 994. [4] B. Tondu, A kinematic model of the upper limb with a clavicle-like link for humanoid robots, International Journal of Humanoid Robotics, Vol.5, No., 8, pp [5] M. Benati et al., Anthropomorphic robotics: I. representing mechanical complexity, Biological Cybernetics, No.8, 98, pp.5-4. [6] S. Caselli, F. Monica and M. Reggiani, Design issues for a mobile manipulator for service tasks, in Proceedings of 6th Int. Workshop on ISSN: ISBN:

8 Proceedings of the th WSEAS International Conference on SYSTEMS Robotics in Alpe-Adria Danube Region, Ljubljana,7, pp [7] G. J. M. Tuijthof, and J. L. Herder, Design, actuation and control of an anthropomorphic robot arm, Mechanism and Machine Theory No.5,, pp [8] J. Echávarri et al., Safety issues for service robots, International Journal of Mechanics and Control, Vol.8, No., 7, pp. -8. [9] H. Iwata, T. Morita and S. Sugano, Human symbiotic humanoid robot with whole-body compliance, in Proceedings of the 4th CISM-IFToMM Symposium, Udine, Italy,, pp [] T. Morita and S. Sugano, Development and evaluation of seven-d.o.f. MIA ARM, in Proceedings of the 997 IEEE International Conference of Robotics and Automation, Albuquerque, New Mexico, 997, pp [] M. Gaggioli, Design and construction of an anthropomorphic robot arm, Master thesis, LARM, University of Cassino, Cassino, Italy, 4. [] Honda, [] WABIAN, [4] H.Gu, M. Ceccarelli, G. Carbone, Design and Operation of -DOF Anthropomorphic Arm for Humanoid Robots, in Proceeding of RAAD8, 7th International Workshop on Robotics in Alpe-Adria-Danube Region, Ancona, 8, paper n.68. [5] N. E. Nava, Design and simulation of a new low-cost easy-operation humanoid robot, PhD thesis, LARM, University of Cassino, Cassino, Italy, 6. [6] N. E. Nava, G. Carbone and M. Ceccarelli, Anthropomorphic design and operation of a new low-cost humanoid robot, in Proceeding of International Conference on Biomedical Robotics and Biomechatronics, Pisa, 6, paper F7. [7] O. Gaudry et al., Design of a manipulator for planetary rovers, in Proceedings of the th CISM-IFToMM Symposium, Gdansk, Poland 994, pp.667. ISSN: ISBN:

Numerical check of a 2DOF transmission for wind turbines

Numerical check of a 2DOF transmission for wind turbines Beibit Shingissov 1, Gani Balbayev 2, Shynar Kurmanalieva 3, Algazy Zhauyt 4, Zhanar Koishybayeva 5 1, 2 Almaty University of Power Engineering