Practical trajectory learning algorithms for robot manipulators

Several alternative learning control algorithms are discussed from both an inverse dynamics and an optimization point of view. The learning laws are derived in discrete time and do not need acceleration measurements. A simple algorithm using a constant learning operator is proposed to run in addition to a simple proportional-derivative feedback controller. Its performance is comparable to other algorithms, and it works under nonideal conditions where the others fail. Two simulation examples on learning dynamic control and learning optimal redundancy resolution are presented