A Method of Collision Prediction by Binocular Velocity Pair under the Condition of Ego and Objects Motion

Abstract

We propose a new method of collision prediction, which is effective in the situation that any object including ego is able to be under the state of motion or stillness. A new concept named as "binocular velocity pair" is introduced in order to achieve a purpose that robot arrives at a goal without coming into collision with a moving or stationary object. The moving distance delta <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">L</sub> on the left camera image and delta <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">R</sub> on the right camera image i.e. "binocular velocity pair" are extracted at each corresponding pixels. Then, the 3D relative direction of the object, the collision location on the robot and the collision time are calculated directly. Simple neural network can execute the calculation. The effectiveness of the proposed method is checked by using a stereo video camera of View Plus Ltd and a moving platform produced by Tokyo Seiki Ltd