Mobile robot navigation using deep learning

Abstract

The ability to navigate is a key aspect of mobile robots, especially so for autonomous ones. However, the presence of glare has always been an issue for them. As these autonomous mobile robots do not have human supervision, they navigate primarily through the use of their onboard cameras for environmental perception and localisation. With the presence of glare, this can seriously hamper their ability to navigate and in the worst case, cause accidents. As such, this paper will attempt to train a machine learning algorithm in order to rectify this issue. The model, when trained, will be able to segment portions of images containing glare, preventing the robot from using them in its navigational duties.