LIDAR based localization using simplified 2D map on agricultural roads

In agricultural environments, there are a lot of changes invoked by several factors such as season, time of day, weather. Therefore, conventional techniques for autonomous driving are not appropriate because they mainly focus on structured urban environments. In this paper, we propose LiDAR based localization method on agricultural roads, using simplified 2D map which contains boundaries of drivable areas and non-drivable areas. Different from detailed accurate 3D maps, the proposed map can be generated and modified easily. We combined the localization method with nonlinear control law for wheeled mobile robots, and applied to an autonomous small truck. Several experiments are performed using the experimental truck in agricultural environments and resulted successful autonomous drive.