Traditionally, autonomous mobile robots utilise active-based sensors such as laser and sonar system for obstacle detection and avoidance. However, many mobile robots are incorporating vision systems for more complex tasks in image recognition and localisation. Visual information obtained from a camera image, and moreover multiple camera images can theoretically provide (under certain conditions) complete 3D structure of a scene. This ultimately makes active sensors redundant, an observation also evident in biological animals. This project looks to develop a visual-based technique for the task for autonomous robots. Students should have skills and interest in developing and implementing computer-vision algorithms, alongside developing control aspects of the mobile robot platform, such as a quadcopter, or ground-based mobile robot, or even a walking robot Spot Mini.

Dr Tobias Low

• School of Engineering

• Artificial Intelligence and Image Processing
• Electrical and Electronic Engineering
• Mechanical Engineering

• Doctor of Philosophy (DPHD)