Bounding Velocity in Tracking Control of Unicycle Mobile Robots with Genetic Programming

This paper introduces a methodology for the synthesis of nonlinear tracking controllers. This approach is applied to the navigation of unicycle mobile robots, where a constrained velocity is pursued. The proposed approach extends the notions of Behavior-based control by redefining the basis behaviors as analytic functions. The conception of natural behavior (which is composed by unforced, forced and learned behaviors) in order to characterize the properties, actions, and restrictions, of the mobile robot, is introduced. Within this approach, the Genetic Programming is dynamically introduced, as a learning process, in the structure of a Control-Theory-based tracking controller. Then, a search of the set of fittest learned behaviors, addressing the complete control problem, is carried out. A selected solution with high fitness value, from the discovered set of learned behaviors, is simulated to show the effectiveness of our proposed framework.