Single Operator, Multiple Robots: Call-Request Handling in Tight-Coordination Tasks

Summary

Many applications of robots require a human operator to supervise and operate multiple robots. In particular, the operator may be required to resolve call requests when robots require assistance. Previous investigations assume that robots are independent of each other, and allow the operator to resolve one request at a time. However, key challenges and opportunities arise when robots work in tightly-coordinating teams. Robots depend on each other, and thus a single failing robot may cause multiple call requests to be issued (by different robots). Moreover, when the operator switches control to a robot, its teammates must often wait idly until the call request is resolved. We contrast previous approaches with two novel distributed methods, where the call-request resolution is itself considered a collaborative problem-solving activity, and non-failing robots use their knowledge of the coordination to assist the operator. We empirically compare the different approaches in several scenarios involving tight coordination, where an operator seeks a dead robot in order to assist it. Extensive experiments with 25 human operators show that this new technique is superior to existing methods, in terms of reducing the time to locate the dead robot. We also show that the new method has much more consistent performance across different operators.

This research was supported in part by BSF Grant #2002401 and by the Israel Ministry of Science and Technology (MOST).