Event
DONIC Dual-Arm Dexterous Tracking Controller and BiHOI Dataset Released
2 reports · 2 sources
The research team proposes DONIC—a general dual-arm dexterous tracking controller jointly trained across manipulation categories and human motion sources—and releases BiHOI, a multi-view dataset containing 1,575 sequences and covering 229 objects, emphasizing sustained dual-arm interaction and retargeting reconstructed human interactions into robot references. The paper reports that, without fine-tuning, a single checkpoint achieves an 86.79% success rate on unseen trajectories of training objects and 57.32% on held-out objects. DONIC combines proximal policy optimization with behavioral cloning into a feedback loop, recycling successfully validated executions as training demonstrations.
Report timeline
- DONIC: Generalist Dexterous Bimanual Tracking Control Learned from Human Demonstrations
DONIC is the first generalist bimanual dexterous tracker jointly trained across manipulation categories and human motion sources. It is released alongside BiHOI, a multi-view dataset containing 1,575 sequences covering 229 objects, in which reconstructed interactions are retargeted into robot references.
arXiv cs.RO ·
- DONIC: Generalist Dexterous Bimanual Tracking Control Learned from Human Demonstrations
The paper presents DONIC, a generalist bimanual dexterous tracker jointly trained across manipulation categories and human motion sources, and releases BiHOI, a multi-view dataset containing 1,575 sequences covering 229 objects, emphasizing sustained bimanual interaction and retargeting reconstructed interactions into robot references.
HuggingFace Daily Papers ·