Robust Optimal Sensor Planning for Occlusion Handling in Dynamic Robotic Environments

Abstract

Optimal sensor planning for workspace detection in robotic environments is hindered due to sensor occlusions. These occlusions are often dynamic. Probabilistic optimization frameworks, which generally deal with the uncertain nature of these occlusions, suffer from unreliability and/or unavailability of probability distribution functions. This paper proposes and analyzes a robust optimization approach (minimax), which generates sensor configurations based on occlusion scenarios that cause maximum obstruction of the robotic workspace. The optimal solution is independent of probability distribution functions and provides a guaranteed level of workspace visibility regardless of occluder positions, thus accounting for random occlusions. The method also allows the user to determine the impact of the worst case occlusion scenarios leading to a broader perspective on sensor planning. Evaluation of the approach for a mobile medical X-ray robotic system in a simulation healthcare environment shows the effectiveness of the proposed method.