RI40 Seminar: Harry Asada : From Direct-Drive to SuperLimb Bionics

In this talk

1. 1:28

   Direct drive and the CMU legacy

   Asada credits CMU with helping him move from theoretical robot dynamics to building a direct-drive arm, which could accurately generate and measure joint torque. He traces the technology to rare-earth magnets and high-power MOSFETs, and describes uses in clean-room wafer handling, where direct-drive robots can achieve positioning accuracy below 20 microns without gear-generated particles.

2. 10:03

   Extra limbs for demanding work

   Asada distinguishes superlimbs from exoskeletons: they add limbs that do not follow the body's usual joint structure. Shoulder-mounted arms could support overhead aircraft assembly, while body-bracing limbs could help workers reach the floor without leaning on their hands. He also describes extra fingers that work alongside a natural hand.

3. 16:41

   Coordinating people and robots

   Asada identifies communication and control as central challenges because wearable robots must coordinate with their users. His lab used demonstrations and data analysis to map a worker's movements to a robot's actions, reducing many degrees of freedom to a few coordination modes. For extra fingers, the team explored controlling them in concert with the natural hand and using small changes in tool grip as commands when both hands were occupied.

4. 28:21

   Human feedback and shared movement

   Asada raises questions about whether people can control extra limbs and come to perceive them as part of their bodies. In experiments, users tended to time-share their attention between their arms and robot limbs, while haptic feedback through the body could help users monitor robot actions without looking at them. He also describes a human-robot walking system where passive springs and dampers helped the two sets of legs settle into a coordinated phase.

5. 43:31

   New actuators for wearable and tiny robots

   Asada presents a dual-motor design that combines a near-direct-drive motor for speed with a geared motor for high torque, addressing tasks that range from moving quickly to supporting a heavy load. He then describes research using living muscle as a robot actuator, including optogenetic control that activates selected muscle strips with light. He sees potential for dense, small-scale robots, while noting the challenges of controlling and coordinating the cells.