紀要論文

3D LiDARの反射強度を用いた自律ロボットのための車線追従アルゴリズムに関する研究

A STUDY ON LANE FOLLOWING ALGORITHM BASED ON 3D LIDAR INTENSITY FOR MOBILE ROBOTS

jpn

24368083

https://doi.org/10.15002/00030771

法政大学大学院理工学研究科

法政大学大学院紀要. 理工学研究科編

65

1

6

2024-03-24

Version of Record

Intelligent Ground Vehicle Competition (IGVC) is an annual competition for autonomous mobile robots held in the U.S. One of the vital functions of this competition is autonomous lane-following tasks that require lane recognition while avoiding obstacles and following within lanes. The most popular approach for lane recognition employs the image sensor such as the camera. However, such an approach essentially has weaknesses due to outdoor lighting condition changes, that are caused by image sensor dynamic range limitations. To overcome the weaknesses of image sensor-based lane recognition, we newly utilize the intensity of the point cloud acquired by 3D LiDAR. The combination of point clouds and intensity enables both to detect lanes and obstacles without the effect of lighting condition change. In this paper, we propose a novel safer, and robust lane-following algorithm that only uses 3D LiDAR without the use of the image sensor. The lane detection is performed by extracting a group of points intensity within a certain range while estimating the direction of the lane and using a threshold value. Detected lanes and obstacles are represented as a one-dimensional cost function, and a local target destination is determined in the direction with the lowest cost, enabling robust control regardless of the outdoor lighting conditions. Lane following, obstacle avoidance, and real-time performance were evaluated in a simulation environment.

Article