Vision-Based Obstacle Avoidance in Sidewalk Environment Using Top-View Transform and Optical-Flow

This paper presents a vision-based obstacle avoidance method for blind pedestrians in an outdoor sidewalk environment.Unlike many existing travel-aid systems using stereo-vision based methods,the proposed method is able to get obstacle position as well as user motion information by just one monocular camera fixed at the belly of the user.To achieve this goal,a top-view transformation of the road image is used for obstacle detection and user motion estimation,based on which a grid map is generated for navigation.For detection part,the bottom points of erect obstacles are detected by extracting local-maxima and minima on the top-view image while user motion is estimated by analysing the optical flow vector field in the user surrounding area.For the obstacle avoidance part,a step score is calculated on the grid map for evaluating the safety of next moving step.Experiments with several sidewalk video-clips show that the proposed obstacle avoidance method is able to provide useful guidance instructions under certain sidewalk environments.