Sensor-based motion-planning of a manipulator to overcome large transmission delays in teleoperation

To overcome the time delay problem, we claim that a slave arm should have an autonomy. In this research, a shared autonomy is designed under the sensor-based motion-planning algorithm. A human roughly selects an approximate sequence of fine motions by a master arm while watching its virtual slave arm in a 3D graphics environment at a master site. Then, a computer at the master site communicates the sequence to a computer at a slave site. In succession, by the sensor-based motion-planning under the impedance control, the communicated sequence is precisely coordinated in a real assembly task at a remote site. Due to the real-time adjustment, a slave arm performs an assembly task such as a peg-in-hole task very well. Furthermore, the sensor-based motion-planning is supported by local sensor information; it is robust against position and orientation errors of a slave arm or communication error between master and slave sites. Finally, the usefulness of the shared autonomy is ascertained in an experiment between master and slave sites.