Edge detection using structured laser pattern and vision for mobile robot navigation

This paper presents a vision-guided navigation strategy based on edge detection. This controlled strategy is computationally simple and suitable for small mobile robots with limited onboard resources, such as a mobile sensor node for structural health monitoring. Traditional vision navigation relies on detecting and extracting features from structured environment or natural scene as references for localization; such approach requires heavy computations for real-time operation and hence results in bulky designs of the mobile robots. To overcome this limitation, a new navigation strategy is proposed by monitoring the change of a projected laser pattern on different obstacles such as edges, walls and corners. Due to the characteristics of laser light, the pattern can be easily captured for fast image processing. Based on the pattern changes, a controlled method for navigation on a plane is designed and the dynamics of a wheeled robot is formulated for simulation. Illustrative applications are provided by results of edge detection using an onboard camera and feedback control for a wheeled robot navigation.