Publications / 2006 Proceedings of the 23rd ISARC, Tokyo, Japan
The problem of coordinating construction vehicles in formations is tackled using the particle swarm optimization algorithm. Coordinated multi-vehicles are anticipated to outperform single vehicles in terms of economy and flexibility. A possible approach is to apply control theoretic methods to vehicle coordination but this may require complicated system models to be employed. The particle swarm optimization (PSO) algorithm adopted in this work, as an evolutionary computation based methodology, is able to provide a near-optimal solution without the need for a precise system model. The control commands for the vehicles are treated as particles in a swarm and a sequence of controls is derived to achieve the desired formation. With regard to inter-vehicle collision avoidances, behaviour-based control strategies are incorporated into the formation framework. Simulations for multi-vehicle formation in a construction site scenario are conducted to illustrate the effectiveness of the proposed approach.