Replanning objectives with transition system equivalencies

A multi-expert conducting system for robotics applications is presented. We identify two types of experts whose conducting approach is guided by specific objectives issued from the global application objectives. We present also the two types of dynamic planning relative to these two experts: replanning the task activation and reprogramming the machines actions. An exhaustive definition of the actions and of universe states of the replanning models is given. We then expose a method for building research spaces, related to the replanned plans and reprogrammed actions, based on the system of transitions equivalencies. We present our approach of replanning for correcting malfunction