Path planning of collision avoidance for unmanned ground vehicles: A nonlinear model predictive control approach. (February 2021)

• Record Type: Journal Article
• Authors: Hang, Peng; Huang, Sunan; Chen, Xinbo; Tan, Kok Kiong
• Is Part Of: Proceedings of the Institution of Mechanical Engineers. Volume 235:Number 2(2021)
• Page Start: 222
• Page End: 236
• Publication Date: 2021-02
• Subjects: Unmanned ground vehicle -- path planning -- collision avoidance -- nonlinear model predictive control; Mechanical engineering -- Periodicals; Automatic control -- Periodicals; Systems engineering -- Periodicals; 621.3
• DOI: 10.1177/0959651820937844
• Languages: English
• ISSNs: 0959-6518

Abstract

In addition to the safety of collision avoidance, the safety of lateral stability is another critical issue for unmanned ground vehicles in the high-speed condition. This article presents an integrated path planning algorithm for unmanned ground vehicles to address the aforementioned two issues. Since visibility graph method is a very practical and effective path planning algorithm, it is used to plan the global collision avoidance path, which can generate the shortest path across the static obstacles from the start point to the final point. To improve the quality of the planned path and avoid uncertain moving obstacles, nonlinear model predictive control is used to optimize the path and conduct second path planning with the consideration of lateral stability. Considering that the moving trajectories of moving obstacles are uncertain, multivariate Gaussian distribution and polynomial fitting are utilized to predict the moving trajectories of moving obstacles. In the collision avoidance algorithm design, a series of constraints are taken into consideration, including the minimum turning radius, safe distance, control constraint, tracking error, etc. Four simulation conditions are carried out to verify the feasibility and accuracy of the comprehensive collision avoidance algorithm. Simulation results indicate that the algorithm can deal with both static and dynamic collision avoidance, and lateral stability.