Humanoid robot workspaces improve by moving cameras with arms
Visible-Reachable Workspace for Perception-Aware Humanoid Design
Robotics
Summary
Robots often need to see what they are reaching for to interact effectively, but just being able to reach something doesn’t mean the robot can see it well. The authors created a new way to measure where a robot can both see and reach targets, called visible-reachable workspace (VRW). They built a humanoid robot with movable cameras to test this idea, showing it could see and grab objects faster and use less energy than a robot with fixed cameras. This approach helps design robots that coordinate movement and perception better.
What this means in practice
- •For robotics engineers: Design humanoid robots with movable cameras to improve simultaneous visibility and reachability of objects for manipulation tasks.
- •For robotics system integrators: Optimize robot configurations to reduce completion time and energy for tasks requiring visual feedback from multiple viewpoints.
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