Research On Planning Algorithms For Autonomous Cleaning Vehicle On Structured Roads

The new round of technological and industrial revolution is in full swing,and intelligent vehicles have become the strategic direction of global automotive industry development.With the rise of autonomous driving technology,autonomous cleaning vehicle emerged as the times require.It is a new generation of cleaning vehicles that can independently perform cleaning operations.The application and implementation of autonomous driving technology in cleaning vehicles can effectively improve the efficiency and quality of urban environmental sanitation work.This paper studies the planning algorithm for autonomous cleaning vehicle in structured road scenarios.According to the characteristics of the cleaning vehicle’s operation and the scene of structured roads,research is conducted on task planning,behavior planning,and motion planning,respectively.The main contents of this paper is as follows:(1)Task planning problem for cleaning vehicles on structured roads.In task planning,this paper converts the road topology connection network into a weighted directed complete graph,thereby transforming the global routing ergodic planning search problem into an ATSP problem.By pruning the lanes at intersections on HD maps,the search efficiency is optimized,and the ergodic path initially obtained by the search is pruned to improve the quality of the ergodic path.Finally,multiple structured map topology connection networks are obtained through OSM format maps to construct the global routing ergodic search problem.The backtracking algorithm,LKH algorithm,and its pruning LKH algorithm are used to solve the problem in the Common Road environment.The experimental results show that the pruning LKH algorithm can meet the task planning requirements for unmanned cleaning vehicles by searching the ergodic path in the weighted directed complete graph.The searched path can achieve 100%coverage of the target area with low repetition coverage rate.(2)A behavior planning scheme for autonomous cleaning vehicle based on behavior trees and finite state machines is constructed.Based on the characteristics of the different motion planning methods required for autonomous cleaning vehicle in different operation scenarios,this paper uses behavior trees for scene recognition and calls different motion planning algorithms according to different scenarios.The finite state machine is used to guide the vehicle to complete the cleaning task in special scenario operation modes.The combination of the two behavior planning methods can not only satisfy the switching of cleaning modes under general structured road conditions and right-angle bending working conditions but also has good scalability for future operation scenarios.(3)A motion planning algorithm for autonomous cleaning vehicle in general cleaning scenarios on structured roads is established.Through analysis of the cleaning vehicle’s operational characteristics and consideration of the driving-along-side cleaning feature,a companion reference line is constructed based on the road centerline and the cleaning vehicle’s parameters.The rough trajectory is generated through the companion reference line,and then the executable trajectory that can perform the driving-along-side cleaning tasks and comply with various constraints is generated through quadratic optimization,considering vehicle performance constraints and road constraints.(4)A motion planning algorithm for cleaning operations in special operation scenarios is established.The trajectory generation algorithm based on the reference line is difficult to perform right-angle bending cleaning operations.Therefore,this paper proposes the Hybrid A*algorithm combined with anchor points to perform ergodic planning in this scenario.Firstly,the right-angle bend is identified,and then the key anchor points are calculated based on the type of right-angle bend and the position of the corner.Finally,the trajectory planned by the Hybrid A* algorithm guides the cleaning vehicle to traverse all the anchor points and complete the cleaning operation under right-angle bending working conditions.