Abstract
Camera vision systems are widely used in autonomous vehicles for position tracking and feedback. The dynamics of an object in the reference frame can be accurately estimated by tracking unique image features. This process is known as Visual Odometry. The motion of the camera or ego-motion is estimated by using the features tracked with visual odometry as reference points of motion. In this project a robust method for computing the translational and rotational ego-motion of a stereo-vision system is proposed.