Data Fusion-Aware Motion Planning for Ad Hoc Robotic Search Teams

This paper develops a novel algorithmic motion planning approach that allows privately-owned volunteer robotic equipment, which might otherwise remain unused, to provide value to a network of relief workers or other robots engaged in a search effort. The specific ‘Volunteer Robot Problem’ considered here is a path planning problem that asks an autonomous volunteer robot to balance information gathering tasks with data fusion when it becomes part of an ad hoc distributed robotic network supporting a deliberate relief effort. Related prior work considered optimal search strategies over information fields, but often these methods assume direct access to high performance centralized computing or to continuous communications for decentralized coordination. In this work, we provide a formal definition for the ‘Volunteer Robot Problem’ and information as it relates to general search tasks, and develop a novel information gathering planning algorithm to solve it. Our method improves upon existing sample-based planning algorithms by accounting for intermittent data fusion opportunities with other search agents, while remaining computationally lightweight and requiring minimal a priori knowledge of both ownship and other agents' states and capabilities. Simulation-based validation and comparisons to alternative planning approaches are provided of the algorithm through simulations for different multi-agent search scenarios and comparisons to other sampling-based algorithms for information-guided path planning.