LAUSR

Mobile manipulation tasks require a seamless integration of navigation and manipulation capabilities. Finding suitable robot placements to pick up and place objects in such tasks is crucial for time-efficient task execution. Sub-optimal robot placements result in infeasible solutions or require larger re-positioning of the mobile base to reach target objects, increasing the overall time to complete the task. In this work, we propose an approach that, given a set of objects, autonomously selects the optimal placements of a humanoid robot in conjunction with the best grasp candidate and corresponding arm. In contrast to previous approaches, our method considers both the navigation costs between consecutive robot placements and the manipulation costs to reduce the time needed to complete the task. We evaluate our method on a simulated table clearing task that requires the robot to move between pickup and discard locations and demonstrate the applicability in a real-world experiment on the humanoid robot ARMAR-6. In addition, we perform a run-time analysis and show that our approach can integrate sensory feedback to update the optimal placement in dynamic environments.