Paper Overview
Field: Robotics / Computer Vision Authors: Naoki Kita, William J. Beksi, Luis Sentis Published: 2026-04-30 arXiv: 2604.28197
Summary
Human-robot collaboration research has focused mainly on dyadic or sequential interaction settings. Real homes, however, require *multiadic* collaboration: multiple humans and robots sharing a workspace and acting concurrently on interleaved subtasks with tight spatial and temporal coupling. This regime remains underexplored because close-proximity interaction among humans, robots, and objects creates persistent occlusion and rapid state changes, making reliable real-time 3D tracking the central bottleneck.
The authors present OmniRobotHome, the first room-scale residential platform that unifies wide-area real-time 3D human and object perception with coordinated multi-robot execution in a single world coordinate frame.
Key contributions
- 48 hardware-synchronized RGB cameras deployed in a natural home environment
- Markerless, occlusion-robust tracking of multiple humans and objects
- Temporal alignment of perception with the real-time scene-state execution of two Franka robotic arms
- Long-horizon human behavior modeling enabled by continuous capture in a consistent coordinate frame, building behavior memory from accumulated trajectories
Focus problems
The platform makes multiadic collaboration experimentally feasible, and the authors focus on two core questions:
1. Safety in shared human-robot environments 2. Anticipatory robot assistance based on human intention/prediction
They show that real-time perception and accumulated behavior memory yield measurable improvements in both.
Original Abstract (excerpt)
> Human-robot collaboration has been studied primarily in dyadic or sequential settings. However, real homes require multiadic collaboration, where multiple humans and robots share a workspace, acting concurrently on interleaved subtasks with tight spatial and temporal coupling. This regime remains underexplored because close-proximity interaction between humans, robots, and objects creates persistent occlusion and rapid state changes, making reliable real-time 3D tracking the central bottleneck. ...
---
Paper: https://arxiv.org/abs/2604.28197