Robot indoor navigation based on computer vision and machine learning

Abstract

Autonomous navigation, as a fundamental problem of intelligent mobile robots’ research, is the key technology of mobile robot to realize autonomous and intelligent. A method of combing computer vision and machine learning for the problem of robot indoor navigation is proposed in the paper. It realizes robot autonomous navigation through imitating the behavior of experts. Through a camera to perceive environmental information, expert provides some examples of navigation for robot to learn and robot learns a control strategy based on these samples using imitation learning algorithm. When robot is running, the control strategy learned can infer a corresponding control command based on the current perception of environmental information. Therefore, robot is able to mimic the behavior of expert to navigate autonomously.