Autonomous mobile robot navigation using machine learning

This paper develops a decision-making system based on the BP Neural Network to navigate a robot in an unknown environment. Based on the neural network model, the robot can move out of specific mazes successfully through adjusting its direction and speed continuously. A BP neural network, which includes three input nodes and nine output nodes, are designed for the navigation system. The information of the surrounding environment is returned by six ultrasonic sensors on the front and bilateral sides of the robot. After thousands of training, the robot learns the navigation knowledge successfully from the samples, and move out of the mazes autonomously. The performance of the robot is validated with the simulation results and two physical experiments. The results show that the robot could navigate autonomously in unknown environments.