Fast planning through planning graph analysis
Artificial Intelligence
CPlan: a constraint programming approach to planning
AAAI '99/IAAI '99 Proceedings of the sixteenth national conference on Artificial intelligence and the eleventh Innovative applications of artificial intelligence conference innovative applications of artificial intelligence
Sokoban: enhancing general single-agent search methods using domain knowledge
Artificial Intelligence - Special issue on heuristic search in artificial intelligence
Planning as constraint satisfaction: solving the planning graph by compiling it into CSP
Artificial Intelligence
Planning Algorithms
Exploiting subgraph structure in multi-robot path planning
Journal of Artificial Intelligence Research
Generalizing GraphPlan by formulating planning as a CSP
IJCAI'03 Proceedings of the 18th international joint conference on Artificial intelligence
Hi-index | 0.00 |
Planning collision-free paths for multiple robots traversing a shared space is a problem that grows combinatorially with the number of robots. The naive centralised approach soon becomes intractable for even a moderate number of robots. Decentralised approaches, such as priority planning, are much faster but lack completeness. Previously I have demonstrated that the search can be significantly reduced by adding a level of abstraction [1]. I first partition the map into subgraphs of particular known structures, such as cliques , halls and rings , and then build abstract plans which describe the transitions of robots between the subgraphs. These plans are constrained by the structural properties of the subgraphs used. When an abstract plan is found, it can easy be resolved into a complete concrete plan without further search. In this paper, I show how this method of planning can be implemented as a constraint satisfaction problem (CSP). Constraint propagation and intelligent search ordering further reduces the size of the search problem and allows us to solve large problems significantly more quickly, as I demonstrate this in a realistic planning problem based on a map of the Patrick Port Brisbane yard. This implementation also opens up opportunities for the application of a number of other search reduction and optimisation techniques, as I will discuss.