Design of a navigation system for a household mobile robot using neural networks

Abstract

A household use mobile robot is expected to perform daily duties such as cleaning and polishing the floor, delivering small items and security patrol. To accomplish these tasks, this robot must navigate in a continuously changing indoor environment and avoid obstacles. A self-navigation algorithm has been developed for such purposes by the use of neural networks. With the aid of self-navigation, this robot can be used to cover the entire floor area for cleaning purposes or move from one place to another for delivery. Through the abilities of learning and competing provided by the neural networks, the robot is capable of exploring unknown environments and searching for optimum paths so that it can fulfil its tasks with smallest overlapping spaces and minimum number of turns for energy conservation.