Experimental evaluation of human-machine collaborative algorithms in planning for multiple UAVs

Many planning applications are theoretically intractable or unable to be solved in a practical amount of time by a human or a computer alone. Typical approaches involve a significant number of heuristics in an effort to deal with the complexity of the problem; limited decision support is provided for a human. An alternative is to integrate a human with the algorithms, so they can become partners. This paper describes an experimental study of such integration using a command and control resource allocation and planning system. The plans created with human involvement in the algorithms are compared to those created by the machine alone. Two levels of human-machine collaboration are studied, one in which the human completes a subproblem independently and the other in which the machine provides support. The results show that improved solutions are achieved when a human is aided by machine support in completing the subproblem.