Multivariable Neuro-Fuzzy Controller for an Autonomous Vehicle Navigation

Resumen

The main function of autonomous vehicle is to carry tasks pre-determined going itself in unknown environments. Therefore, the vehicle needs a system both to control a task planning and to guarantee a decision taking development, in real time during the displacement, carrying the tasks and avoiding potential obstacles. This work proposes a multivariable neuro-fuzzy controller implementation for autonomous vehicle navigation. The controller integrates the systems that determine the tasks and guarantee the displacement with security. To prove the purpose controller performance it is made a comparison with presented neuro-fuzzy controller in literature. The verification of multivariable neuro-fuzzy controller performance is gotten through simulations in different environments with obstacles. The software MatLab 6.5 simulations, proving that in the case of complex systems, for example an autonomous vehicle, the architecture of propose controller is less complex and with better performance in relation to neuro-fuzzy controller.