Humanoid robot design improves reach and camera vision for better tasks
Visible-Reachable Workspace for Perception-Aware Humanoid Design
Robotics
Summary
Robots that pick up and move things usually measure where their arms can reach, but they often forget whether they can see those spots well enough to do the job. The authors came up with a way to measure not just reach but also what parts of the workspace a robot's cameras can see while reaching. They built a humanoid robot with movable cameras and showed it can see and reach much more of its workspace compared to fixed cameras. This means the robot can work faster and use less energy, and it can look and grab targets on different sides without moving its whole body.
workspace analysishumanoid robotend effectorRGB-D cameravisible-reachable workspacekinematic reachabilityperception-aware designactuated camerasrobot manipulationspatial coverage
Authors
Boxi Xia, Zijiang Yang, Ryan Shin, Bokuan Li, Eric Wun-Hao Lu, Jacob Lee, Jiaxun Liu, Boyuan Chen
Abstract
Workspace analysis measures where a robot can place its end effector. For visually guided manipulation, reachability alone is insufficient: a kinematically reachable target may not be visible in the specific pose required to reach it. The robot must then redirect its sensing or move its body to acquire a view, turning a perception limitation into additional motion. Existing humanoids largely inherit this limitation when copying human form factors. We introduce the visible-reachable workspace (VRW), a design-stage measure that conditions visibility on feasible reaching configurations and extends it to concurrent visibility of spatially separated work regions. We apply VRW by building a 31-DoF humanoid with independently actuated RGB-D cameras. On the same robot, camera articulation increases visible-reachable coverage from 38% to 97%. With actuated camera layouts, a second camera raises pairwise coverage from 0.45 to 0.95, while a third changes it only to 0.97. In a controlled two-target reach-and-grasp benchmark, our dual-actuated design reduces mean completion time by 17% and mechanical energy by 19% relative to the same robot with its cameras fixed. Hardware experiments demonstrate simultaneous observation and manipulation of front/back and left/right target pairs without torso reorientation. The results suggest that reachability becomes a more informative design quantity for perception-driven humanoid manipulation when it is evaluated together with the sensing configurations that make the reachable space observable. We will open-source all software and the humanoid hardware design. Our website is https://generalroboticslab.com/DukeHumanoidv2