Find a Multi-agent planning solution in nondeterministic domain

In the Multi-agent planning domain, every agents have its own target state, and the planning problem is that the agents arrive the target state from the initial state without conflict. Finding a Multi-agent planning solution in the deterministic domain is a very hot problem. However, it is a problem to be studied in the nondeterministic domain. We study on the issue, and propose a solution algorithm. Firstly, the idea of the algorithm is to take the state layered by the hierarchical thinking. Secondly, search by the partial greedy choice and backtracking according to the conflicting list of the current state set of the agents. Finally, find a collaborative planning without conflict.