Neuro-fuzzy based system for autonomous vehicle parking in the dynamic environment

Abstract

This paper proposes a multi-stage, neuro-fuzzy architecture for autonomous parallel parking in a complex and dynamic environment. It provides an obstacle avoidance capability for the vehicle during the parking maneuver. Fuzzy controller transforms input information into effective vehicle parking by switching between navigation and parking modules. In order to sense the environment better, a trained neural network is appended as an input pre controller to the central fuzzy controller for obstacle avoidance. To demonstrate the efficacy of the proposed architecture, simulation tests are carried out in the presence of both static and moving obstacles.