9318496 Mason The goal of this research is to enable robotic manipulators to make better use of dynamic phenomena. At present, most practical robotic manipulation is dominated by a kinematic approach. Typically an object is rigidly grasped by or affixed to the end effector, and the manipulator is moved along a specified trajectory without explicitly involving analysis of the dynamics of the system. A dynamic approach to manipulation extends an arm's capabilities in several ways. The use of motions tuned to the dynamics increases the speed and maximum load. The ability to achieve object motions within the hand effectively increases the motion freedoms of the system. Dynamic methods can have a broad impact on robotic manipulation, with particular applications in manufacturing and space construction. The approach of this research is to explore task domains in which impact, momentum, and dynamic forces are dominant effects, such as in catching and throwing of objects. While these operations were chosen to present significant challenges to manipulation, more than for any immediate practical value, the ability to throw or drop objects may in some cases extend the effective workspace of the manipulator. This exploration will involve construction of simple manipulators, development of methods to analyze dynamic manipulation tasks, and production of control and planning systems. What is learned here will provide important new knowledge and approaches to dynamics of manipulation applicable to many problem domains.
