Real-time Robot Placement-based Inverse Reachability Map

This paper presents an efficient method for determining optimal robot base placement based on target end-effector poses. While traditional approaches involve time-consuming online computations, our method uses a precomputed 3D inverse reachability map (IRM) to reduce computational costs. The 3D IRM is constructed offline by dividing the workspace into voxels and sampling potential base poses relative to each end-effector pose. Given a specific task, a 2D IRM is generated by slicing the 3D IRM with a task-specific plane and transforming it into the world frame. This enables rapid evaluation of candidate base poses on a predefined ground plane. We further integrate multiple 2D IRMs for different task poses and evaluate candidate base configurations using an accumulative manipulability score derived from inverse kinematics solutions and manipulability measures. Our algorithm significantly reduces computation time compared to other methods while ensuring feasible and manipulable base placements for the task poses.