Inter-vehicle separation measurement using monocular vision

Inter-vehicle separation measurement is crucial for intelligent vehicle system, in particular the autonomous vehicle. The measurement results can be used for regulating a safe inter-vehicle distance when an autonomous vehicle is trailing behind a leading vehicle. Furthermore, the measurement results can also be used for planning a safe path for over taking another vehicle by the autonomous vehicle On the other hand, vision-based vehicle localization strategy has been widely used to acquire the depth information of vehicles ahead of the ego vehicle. This paper proposes an improved method of inter-vehicle separation measurement, between the ego vehicle and the lead vehicle, using a monocular vision system. As the effect of traction and interaction between the wheels of the vehicle and the road surfaces, pitching effect is inevitable for the autonomous vehicle. As the pitching angle of the camera, mounted on the autonomous vehicle, is a crucial factor affecting the precision of the inter-vehicle separation measurement, this paper attempts to improve the parallel lane lines method (PLLM) for this purpose. An angle sensor has been introduced to accurately acquire the pitching angle of the camera that is mounted onto the ego vehicle. As compared to the PLLM, the introduction of the angle sensor has resulted in an accurate and reliable pitching angle compensation for the camera, thus satisfying the requirement for real time accurate inter-vehicle separation measurement. The proposed solution has also resulted in robust autonomous vehicle that can be deployed in any road conditions in the urban city environment. Furthermore, the proposed solution is less computation intensive and hence improving the efficiency of the autonomous vehicle system. Experiments have been conducted to test the performance of the two methods, which verify the advantages of the introduction of angle sensor for pitch angle compensation for the monocular vision system.