Reflexive Control of a Rover for Planetary Exploration with Artificial Neural Networks

摘要

In this paper a Reflexive control layer for an autonomous exploration rover based on artificial neural networks is proposed. It has to control the motors of the rover following the requests of a high level system, which is responsible of the execution of a plan, and to monitor the health status of the rover at low level. The core of the reflexive layer is a controller which has been designed using an artificial neural network. The structure of this controller is reported in detail, together with the training process adopted and the preliminary results obtained applying the proposed controller to a six wheeled rover.