Design of dynamic obstacle avoidance system for self-driving sweeper based on lattice-planner

Abstract

In this paper, a dynamic obstacle avoidance control system of self-driving sweeper is designed based on lattice-planner algorithm in robot operating system (ROS). On the basis traditional manual sweeper, an obstacle avoidance method in line with the driving characteristics of sweeper has been proposed. Firstly, the planning starting point of this frame is calculated according to the mileage points of the previous one. The summary information of static and dynamic obstacles is obtained according to the point cloud data of lidar. And then the horizontal and vertical trajectories are generated in accordance of ST and SL diagrams. After the feasibility analysis of the trajectory, a large number of trajectory bundles are eliminated. The cost detection is carried out for the trajectory bundles that meet the rules, and the collision detection is carried out for the only few remaining trajectories. Finally, through real vehicle debugging, the scientificity and safety of the algorithm in application are verified.