MGRO Recognition Algorithm-Based Artificial Potential Field for Mobile Robot Navigation

Abstract

This paper describes a novel recognition algorithm which includes mean filter, Gaussian filter, Retinex enhancement method, and Ostu threshold segmentation method (MGRO) for the navigation of mobile robots with visual sensors. The approach includes obstacle visual recognition and navigation path planning. In the first part, a three-stage method for obstacle visual recognition is constructed. Stage 1 combines mean filtering and Gaussian filtering to remove random noise and Gauss noise in the environmental image. Stage 2 increases image contrast by using the Retinex enhancement method. Stage 3 uses the Ostu threshold segmentation method to achieve obstacle feature extraction. A navigation method based on the auxiliary visual information is constructed in the second part. The method is based on the artificial potential field (APF) method and is able to avoid falling into local minimum by changing the repulsion field function. Experimental results confirm that obstacle features can be extracted accurately and the mobile robot can avoid obstacles safely and arrive at target positions correctly.