analiza aiqbench
This research introduces BRIDGE, a gating mechanism that intelligently routes between handheld and teleoperated data sources to improve robot manipulation in contact-rich scenarios. By selectively applying high-fidelity teleoperated demonstrations only during sensitive task phases, the method achieves significant performance gains over standard handheld-only training baselines. The work advances imitation learning by resolving the trade-off between scalable data collection and precise action supervision without requiring a robot-in-the-loop for entire tasks.
Bridging Handheld and Teleoperated Supervision for Contact-Rich Manipulation via State-Gated Experts
arXiv:2606.26603v2 Announce Type: replace Abstract: Handheld data collection systems, such as the Universal Manipulation Interface (UMI), enable scalable data collection across diverse environments but only capture observed actions rather than the desired actions executed by a robot controller. In contrast, teleoperation captures desired actions directly, but requires a robot in-the-loop. We revisit this trade-off through the lens of action validity across task phases. We observe that handheld trajectories provide valid supervision in tolerant, free-space phases, but can be poor supervision targets for contact-sensitive phases, requiring high stiffness to minimize tracking errors, resulting in large contact forces. We study the interaction between these two supervision types for contact-rich manipulation and find that training policies that combine handheld data with a small number of targeted teleoperated demonstrations provide an effective hybrid strategy. Specifically, rather than teleoperating the entire task, we only collect partial teleoperated demonstrations for task segments where base handheld policies fail. However, naively mixing handheld and teleoperated phase-specific data yields worse performance than training on handheld data alone. To address this mismatch between observed and desired supervision, we propose Bi-modal Routing for Imitation Data via Gated Experts (BRIDGE), a mixture of diffusion policy experts that routes between specialist task phase heads conditioned on the current robot state. Notably, our approach enables task-phase specific use of desired actions during contact sensitive segments and improves success rates over handheld-only baselines by up to 36.7% across three contact-rich manipulation tasks.
Abstract:Handheld data collection systems, such as the Universal Manipulation Interface (UMI), enable scalable data collection across diverse environments but only capture observed actions rather than the desired actions executed by a robot controller. In contrast, teleoperation captures desired actions directly, but requires a robot in-the-loop. We revisit this trade-off through the lens of action validity across task phases. We observe that handheld trajectories provide valid supervision in tolerant, free-space phases, but can be poor supervision targets for contact-sensitive phases, requiring high stiffness to minimize tracking errors, resulting in large contact forces. We study the interaction between these two supervision types for contact-rich manipulation and find that training policies that combine handheld data with a small number of targeted teleoperated demonstrations provide an effective hybrid strategy. Specifically, rather than teleoperating the entire task, we only collect partial teleoperated demonstrations for task segments where base handheld policies fail. However, naively mixing handheld and teleoperated phase-specific data yields worse performance than training on handheld data alone. To address this mismatch between observed and desired supervision, we propose Bi-modal Routing for Imitation Data via Gated Experts (BRIDGE), a mixture of diffusion policy experts that routes between specialist task phase heads conditioned on the current robot state. Notably, our approach enables task-phase specific use of desired actions during contact sensitive segments and improves success rates over handheld-only baselines by up to 36.7% across three contact-rich manipulation tasks.
Robotics (cs.RO)
arXiv:2606.26603 [cs.RO]
(or arXiv:2606.26603v2 [cs.RO] for this version)
https://doi.org/10.48550/arXiv.2606.26603 Focus to learn more arXiv-issued DOI via DataCite
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Text integral sub licența deschisă a publicației · 323 cuvinte
Subiecte conexe
Compliance for Free: Learning Identifiable Impedance via Bilateral Teleoperation
arXiv · 18 sept. 2026
HACo: Learning Haptic Active Compliance for Force-Aware Dexterous Manipulation
arXiv · 30 sept. 2026
SkillWeave: Weaving Heterogeneous Demonstrations into Long-Horizon Manipulation Skills
arXiv · 9 oct. 2026
VibeAct: Vibration to Actions for Contact-Rich Reactive Robot Dexterity
arXiv · 24 sept. 2026