Dallin Swiss and Dr. Mark Colton, Mechanical Engineering
Introduction
A group of BYU researchers and therapists designed a robot named “Troy” for use in studies to investigate robot-assisted therapy for children with autism [1]. Troy was considered effective and successful but studies suggest that developing a new robot with greater realism and additional capabilities may promote greater clinical benefits [2]. The next robot must be safer and enable physical interaction between the robot and child. This can be done by controlling the weight, speed and flexibility of the robot arms. In particular, reducing their weight would reduce the inertial forces put on the child when interacting with the robot
The arm joints on Troy were driven directly by servo motors. Alternatives to a direct connection, such as pulley systems and Bowden Cables (BCs), have been used in other robots such as BioRob [3] and Probo [4], [5]. These alternatives allow for relocation of the servos/motors and would reduce the weight of the arms. But while the distance between pulleys must remain fixed, BCs can change their position and direction and still transmit motion. BCs require only that the housing ends remain fixed with respect to the servo and the joint. This report is focused on exploring the efficiency of these three actuation methods (direct actuation, pulley and cable actuation, and Bowden Cable actuation) for future use in the design of a robot arm.
Methodology
The three actuation methods were compared on the basis of the amount of electrical power required to perform the same task. Each actuation method was given the task of rotating a 13.3” x 3.5” x 0.125” steel beam (to simulate a robot arm) 5° against the resistance of a spring. See Figure 1 for the direct actuation concept. The other methods followed the same conceptual design and varied only slightly in their physical setup. For the pulley and BC configurations, the servo was set 6” back from the beam and was connected by two small stranded wires. Both wires wrapped once around a pulley at the servo (see Figure 2), and again around a pulley under the beam. The wires were fastened to the beam and were pulled tight using a pair of turnbuckles (see Figure 3). The servo and beam were designed to rotate the same amount in a one-to-one fashion by transmitting torque through small steel pins (see Figure 2). Five different experiments are presented: 1. Direct, 2. Pulley, 3. BCs Straight, 4. BCs Small Turn, 5. BCs 180° Turn. Each experiment consisted of 30 5° movements where electrical current and voltage were sampled at 100 Hz using a portable data acquisition system.
Results
The following table provides data about the average levels of electrical current and voltage for each actuation method. From these, power was calculated and compared to evaluate efficiency.
Discussion
Servos have an internal controller that draws the proportional amount of power needed to reach each new position. The closer it is to the next position, the less power it draws. Somewhere between the start and end angles, there is a point where the servo torque drops to the reaction torque of the spring. This power plateau is different for each actuation method due to friction, housing length, turn radius, and other factors. Consider the data in the results table. Direct required an average of 1.169 W, the least amount of power as expected. In comparison to Direct, the efficiencies of Pulley and BCs Straight are considerably less at 54% and 65% respectively. Their efficiencies are similar since their cable lengths and pulley distances are identical, but further investigation would be necessary to determine why BCs Straight slightly outperforms Pulley. Further work might also reveal why the standard deviation for BC methods is higher than Direct or Pulley. BCs Small Turn suffers from possible pinching in its tight corners putting its efficiency down to 24%. BCs 180° Turn is the least efficient at 21% suffering from friction along the long cable housing. It is coincidental that the two efficiencies just mentioned are similar, but it suggests that a more efficient configuration exists between their extremes. In other words, BCs could be more efficient by reducing the housing length and increasing the turn radius.
Conclusion
Direct actuation requires the least power to move a joint. However, at the cost of efficiency, BCs can provide flexibility in the placement of a servo/motor. If BCs are selected, the driving servo/motor must have at least 4-5 times more torque capability.
Acknowledgements
We thank ORCA for helping fund this research. We also thank members of Mark Colton’s research lab for their involvement in supporting this work.
Scholarly Sources
- M. A. Goodrich, M. B. Colton, B. Brinton, M. Fujuki, J. A. Atherton, and L. Robinson, “Incorporating a robot into an autism therapy team,” IEEE Intell. Syst., pp. 1541–1672, 2012.
- D. J. Ricks and M. B. Colton, “Trends and considerations in robot-assisted autism therapy,” in 2010 IEEE International Conference on Robotics and Automation, 2010, pp. 4354–4359.
- T. Lens and O. von Stryk, “Design and Dynamics Model of a Lightweight Series Elastic Tendon-Driven Robot Arm,” in 2013 IEEE International Conference on Robitics and Automation (ICRA), 2013, pp. 4497–4503.
- G. Van De Perre, R. Simut, B. Vanderborght, J. Saldien, and D. Lefeber, “About the design of the social robot Probo, facilitator for ASD therapies,” in 9th National Congress on Theoretical and Applied Mechanics, Brussels, 9-10-11 May 2012, 2012, no. May, pp. 1–6.
- K. Goris, J. Saldien, B. Vanderborght, and D. Lefeber, “How to achieve the huggable behavior of the social robot Probo? A reflection on the actuators,” Mechatronics, vol. 21, no. 3, pp. 490–500, Apr. 2011.