Motion Planning for Underactuated Mechanical Systems
We are
studying "underactuated" robot systems --- robots with fewer
actuators than degrees-of-freedom to be controlled. Our purpose is to build inexpensive, mechanically simple
robots, transferring the complexity of a system to its design and control. For example, a robot manipulator with
just one or two actuators can control the motions of parts with more
degrees-of-freedom, performing purposeful manipulation by pushing, rolling,
shaking, juggling, and throwing and catching.
This
talk will focus on our recent work on trajectory planning for a class of
underactuated mechanical systems called "kinematically controllable"
systems. I will describe this new
notion of controllability for second-order mechanical control systems,
demonstrate how this property simplifies the motion planning problem, and
present implementations of the theory on an underactuated robot arm and an underactuated
vehicle.
Friday, April 4, 2003
3:30 – 4:30 p.m.