Multi-Resolution Informative Path Planning for Small Teams of Robots

Abstract

Unmanned aerial vehicles can increase the efficiency of information gathering applications. A key challenge is balancing the search across multiple locations of varying importance while determining the best sensing altitude, given each agent's finite operation time. In this work, we present a multi-resolution informative path planning approach for small teams of unmanned aerial vehicles. We model our problem as a team orienteering problem, aiming to maximize reward by performing searches over a set of spatially separated regions. We convert each region into a set of nodes across multiple fixed altitudes, and compute a cost and reward for each node based on sensing resolution at discrete altitudes. We utilize a linearization method to precisely capture the nonlinear information gain reward for each node, which allows us to leverage mixed-integer linear programming optimizers to solve our problem. Through this approach, we’re able to generate plans for our team of agents that balance revisiting regions of importance and exploring new regions. We evaluate our approach against greedy, naive greedy, and random baselines for teams of up to three agents on multiple maps with varying information distributions. We show that our approach can produce plans of greater optimality within a fixed time limit and limited sensing budget over the baselines. We also discuss the tradeoffs in solution quality and runtime over the optimization process compared to the baseline solutions.