Environment Recognition with FMCW-LiDAR-based Dynamic Occupancy Grid Mapping

In automated driving systems, it is still a challenging task to accurately recognize position, shape and velocity of moving and stationary objects surrounding the ego-vehicle. To address this task, dynamic occupancy grid map (D-OGM), which is the expansion of conventional static occupancy grid map (OGM) to dynamic environments, has been researched in recent years. The D-OGM has the capability to recognize obstacles with their position, shape and velocity for both moving and stationary objects. However, there are still challenges that remain for the D-OGM regarding velocity estimation. In this paper, we focus on the problem of the D-OGM where velocity is incorrectly estimated for static obstacles that have elongated shapes, such as guardrails, and misinterpret them as moving objects. To overcome this problem, we develop the D-OGM algorithm incorporating Doppler velocity measurements with FMCW-LiDAR. In addition, an idea to mitigate the degradation on the velocity estimation caused by the micro-Doppler effect, which specifically occurs in FMCW-type sensors, is proposed. Finally, we present the evaluation results of these algorithms in several test scenarios using real sensor data.