Replanning of Multiple Autonomous Vehicles in Material Handling

The fully automated docks in Australia present opportunities for applications of autonomous vehicles and engineering innovation. When planning tasks to be done by multi-autonomous vehicles in an enclosed area with a known dynamic map (i.e. bi-directional path network), there are many issues that have not yet been comprehensively solved. The real world presents more complexity than the initial algorithms addressed. There are problems that occur due to interaction with the real-world. This means autonomous vehicles can stop, are affected, or face problems, and hence tasks and vehicles´ paths and motion need to be replanned. In order to replan, a greater understanding of the state of vehicles, the state of the map, and importantly the importance of tasks and vehicles is definitely needed. This paper explores the improvements made to replanning by gaining a thorough understanding of the map and then utilising map information to make the best, most efficient replanning decision. Five replanning methods are investigated and four options which combine the methods in different ways are tested in this research. A map analysis method is also presented. Simulation studies show that map information based replanning is the most efficient method out of those tested