Abstract

The paper develops a fairly general framework for derivation of control strategies applying to moving vehicles, such as mobile robots or robot arms, in a dynamically changing environment. The basic idea is to consider stochastic terms for any uncertain future environment change and to apply stochastic dynamic programming for control strategy development. The method permits consideration of a number of possible missions such as obstacle avoidance or obstacle hiting and/or moving forward or following a trajectory etc. A number of examples demonstrate that the developed control strategies are capable of taking efficient, human intelligence like decisions in quite complicated conflict situations.