Better remote control of humanoid robots
The EU-funded training project H2R (Bringing human neuromotor intelligence to robots) increased fundamental understanding of human neuromotor control and exploited it to optimise robot performance. The ultimate goal was improved control strategies for teleoperation of dual-arm robots in which a human operator remotely controls the motion. The ability to jointly optimise the 'intelligence' of a remote human operator and of the robot itself will extend performance capabilities and open the door to exciting new applications. Scientists developed a task-space hybrid adaptive mechanism to move a robot arm and end-effector in response to an external disturbance at any place in the robotic system. The mechanism exploited both force and impedance sensors and a fuzzy logic-based scheme to select the best tuning gain. The gain, an increase (positive) or decrease (negative) in output as a result of a feedback control algorithm, is critically important for fine adaptive control of the robot arm and end-effector. In parallel, the team created new knowledge regarding multiple-arm coordination in the context of multi-agent systems, including theoretical studies of uncertain couplings. Scientists also investigated hand gesture-based control in teleoperation of humanoid robots (iCub and Baxter). Work employed two commercially available haptic or 3D touch devices (the PHANTOM Omni and the Novint Falcon) for force feedback. H2R has contributed important increased understanding of the technical requirements for optimal teleoperation of a human-robot system. The team has delivered advanced mechanisms relying on pattern recognition and adaptive gain tuning that attempt to match the telerobot's controller with the human operator's motion pattern. Excellent collaboration with other Chinese institutes and with universities in the United Kingdom has ensured that knowledge transfer will pollinate new ideas and discoveries beyond the project's duration.
Keywords
Humanoid, robots, dual-arm robot, neuromotor, teleoperation