A Novel Three-layer-architecture based Planning Method and Its Applications for Multi-heterogeneous Autonomous Land Vehicles

摘要

This paper presents a novel Three-layer-architecture based planning method for heterogeneous Au-tonomous Land Vehicles (ALVs). To improve the planning problem for multi-robot systems, especially for heterogeneous ALVs, a novel Three-layer-architecture, including a Global-road-network layer, a Path-trajectory layer, and a Planning-executive layer, is introduced. The Global-road-network layer includes real road-network information and algorithms for global planning according to specific applications. The Path-trajectory layer mainly reflects the different motion characteristics of different ALVs generated by actual driving or by simulation. The Planning-executive layer implements the local planning algorithm for avoiding obstacles, which is closely related to the Path-trajectory layer. Several typical tasks by using multiple heterogeneous ALVs are employed to verify the proposed Three-layer-architecture, such as multiple ALVs encircling a single target, multiple ALVs tracking multiple targets, and multiple ALVs to multiple targets patrolling tasks. Some classical algorithms and their improved forms are also employed in these corresponding layers to complete the plan-ning tasks. Experimental results show that the performance of the proposed algorithm is robust and stable.