Stereo vision based real-time obstacle distance and dimension estimation using look-up table

In this study, a system, which enables autonomous vehicles or mobile robots to detect and avoid obstacles in real-time using two active cameras in dynamic environments, was developed. System is tested in three different scenarios. In the first scenario, the platform which cameras are entagrated to is stable and the objects in the environment are mobile. In the second scenario, the platform which cameras are entagrated to is mobile and the objects in the environment are stable. In the third scenario, both the platform which cameras are entagrated to and the objects in the environment are mobile. In many studies, the distance between obstacles and camera is estimated using distance between cameras, focal length and disparity value. In these methods, the fault ratio, while estimating the distance betwwen cameras and objects, is high when especially the objects are far away from cameras and have low disparity values. In this project, a new approach is proposed in estimating the distance between camera and obstacles. In this approach, a look-up table consists stereo distance values matching with real world distance values is used. In the case of obtaining correct disparity map, the distance between the obstacle and the camera, and the dimensions of the obstacle are estimated correctly with the ±10 cm tolerance.