Reactive mobile manipulator base sequence planning for inspection in dynamic environments

This paper presents an optimization-based methodology for a mobile manipulator base sequence planning, with a focus on multi-target inspection tasks. The proposed method exploits an ellipsoid-based model of the robot's reachable space to maximize the number of reachable targets (defined by a set of AprilTags) from each mobile base pose, while minimizing collision risks and base repositioning operations. The proposed optimization approach incorporates dynamic obstacle perceptionusing an RGB-D camera to determine whether an obstacle is static or dynamic, and to react accordingly. The performance of the proposed method is demonstrated through experiments in a digital twin of a laboratory scenario with realistic constraints.