Supermodular Optimization for Redundant Robot Assignment under Travel-Time Uncertainty
share
This paper considers the assignment of multiple mobile robots to goal locations under uncertain travel time estimates. Our aim is to produce optimal assignments, such that the average waiting time at destinations is minimized. Our premise is that time is the most valuable asset in the system. Hence, we make use of redundant robots to counter the effect of uncertainty. Since solving the redundant assignment problem is strongly NP-hard, we exploit structural properties of our problem to propose a polynomial-time, near-optimal solution. We demonstrate that our problem can be reduced to minimizing a supermodular cost function subject to a matroid constraint. This allows us to develop a greedy algorithm, for which we derive sub-optimality bounds. A comparison with the baseline non-redundant assignment shows that redundant assignment reduces the waiting time at goals, and that this performance gap increases as noise increases. Finally, we evaluate our method on a mobility data set (specifying vehicle availability and passenger requests), recorded in the area of Manhattan, New York. Our algorithm performs in real-time, and reduces passenger waiting times when travel times are uncertain.
READ FULL TEXT