Kinect depth sensor for computer vision applications in autonomous vehicles

Abstract

Unmanned vehicles (Smart cars, drones, robots) need to understand and respond to the surrounding environment in order to perform their tasks. Therefore, they must be equipped with vision capabilities (vision sensors) to let them to detect the presence of objects in their paths or measure how far away objects are from the sensor depending of the type of application. Several technologies have been used to deal with this computer vision application such stereovision, LIDAR and RADAR but each of them still has advantages and disadvantages in term of limitations and price. In this paper, we developed one of the functions of unmanned vehicles to detect the presence of objects and measure how far they are using Kinect depth sensor, a low-cost range sensor along with its higher depth fidelity and attractive alternative in computer vision. The results showed that Kinect sensor can detect along with segmentation techniques the presence of objects in its field of view and measure how far they are from the sensor. Based on results, Kinect can be mounted on unmanned vehicles like a vision sensor for obstacle avoidance application or other applications which require a vision sensor to measure how far detected objects are from it as well for also manned vehicles to alert drivers but mostly computer vision has a crucial role in vehicles without human intervention.