Obstacle Avoidance for Autonomous Driving using Neuro-Fuzzy Architecture in an Urban Landscape

Abstract

An autonomous vehicle must be able to respond to the state of objects in the surrounding, be it stationary or in motion. This paper outlines the techniques to allow the car to be aware of its immediate surroundings in front of it while it is moving autonomously and to make decision on its next course of action. If the object in front is stationary or moving slowly, the autonomous vehicle could decide to go around it, provided there are no obstacles at the side of the object in front. If it is unable to go around it, it should stop within a safe distance from the object in front. Several sensors are used to ascertain the proximity to surrounding front objects to make decision on the next movement. One of the techniques used is to apply fuzzy logic on the ultrasonic sensor readings to gauge the distance to front surrounding objects in fuzzy terms such as near, medium-distance or far. This method complements the trained neural network in making decisions on obstacle avoidance.