Overview of Optical Flow Technique for Mobile Robot Obstacle Avoidance

Abstract

Visual navigation is a recent trend to navigate robot. This paper presents the utilization of optical flow technique to achieve this with robust results. It helps in detecting the relativistic relation and related movement of the object. An important task for this is, to detect the obstacles and execute navigation without colliding with an obstacle when the robot is moving. Optical flow is calculated from the sequence of images in the video grabbed by the camera raised on the robot. From this, we further calculate the focus of expansion (FOE). This helps to find the time to contact (TTC). This calculation will provide information about an obstacle in the path, depth and the time needed to collide with the obstacle. By this information, the robot can navigate autonomously without the interference of human, since this all is done onboard. For implementing this, we will use coarse to fine warping method. An experiment will be performed under various environment settings, which will show fewer errors than the earlier implementation for mobile robot and increased performance under noise.