Motion state estimation of surrounding objects using Dynamic Occupancy Grid Map
Daisuke SARUYAMA, Naoki Suganuma, Keisuke Yoneda · The Proceedings of JSME annual Conference on Robotics and Mechatronics (Robomec) · 2019
Autonomous vehicles have been developed to reduce traffic accidents and use as means of transporting the elderly. Autonomous vehicles require surrounding environment recognition with LIDAR to detect obstacles. In the conventional method, if dynamic obstacles are detected, dynamic state is estimated using Kalman filter, and if static obstacles are detected, map of static environment is made. However, this method sometimes cannot be consistent between dynamic obstacles and static obstacles. Therefore, this paper introduce Dynamic Occupancy Grid Map using LIDAR information. In Dynamic Occupancy Grid Map, the environment is represented by 2D grid, and this method is able to estimate velocity and object existence probability on each grid, so it is possible to recognize dynamic obstacles and static obstacles at the same time. In this paper, after making the map using actual driving data, we consider the estimation results of dynamic objects and static objects.