NASA NTRS1987
A fundamental problem which must be resolved in virtually all non-trivial robotic operations is the well-known inverse kinematic question. More specifically, most of the tasks which robots are called upon to perform are specified in Cartesian (x,y,z) space, such as simple tracking along one or more straight line paths or following a specified surfacer with compliant force sensors and/or visual feedback. In all cases, control is actually implemented through coordinated motion of the various links which comprise the manipulator; i.e., in link space. As a consequence, the control computer of every sophisticated anthropomorphic robot must contain provisions for solving the inverse kinematic problem which, in the case of simple, non-redundant position control, involves the determination of the first three link angles, theta sub 1, theta sub 2, and theta sub 3, which produce a desired wrist origin position P sub xw, P sub yw, and P sub zw at the end of link 3 relative to some fixed base frame. Researchers outline a new inverse kinematic solution and demonstrate its potential via some recent computer simulations. They also compare it to current inverse kinematic methods and outline some of the remaining problems which will be addressed in order to render it fully operational. Also discussed are a number of practical consequences of this technique beyond its obvious use in solving the inverse kinematic question.