A neural network supervisor for behavioral primitives of autonomous systems

The authors present a neural network implementation of a fusion supervisor of primitive behavior to execute more complex robot behavior. The neural network implementation is part of an architecture for the execution of mobile robot tasks, which is composed of several primitive behaviors, in a simultaneous or concurrent fashion. The architecture allows for learning to take place. At the execution level, it incorporates the experience gained in executing primitive behavior as well as the overall task. The neural network has been trained to supervise the relative contributions of the various primitive robot behaviors to execute a given task. The neural network implementation has been tested within OPMOR, a simulation environment for mobile robots, and several results are presented. The performance of the neural network is adequate