PrioriTouch: Adapting to User Contact Preferences for Whole-Arm Physical Human-Robot Interaction

Rishabh Madan, Jiawei Lin, Mahika Goel, Amber Li, Angchen Xie, Xiaoyu Liang, Marcus Lee, Justin Guo, Pranav N. Thakkar, Rohan Banerjee, Jose Barreiros, Kate Tsui, Tom Silver, Tapomayukh Bhattacharjee
Proceedings of The 9th Conference on Robot Learning, PMLR 305:1515-1530, 2025.

Abstract

Many robot caregiving tasks, such as bathing, dressing, and transferring, require a robot arm to make contact with a human body at multiple points rather than solely at the end effector. However, varied human touch preferences can lead to unsafe or uncomfortable multi-contact interactions. To address this, we introduce PrioriTouch, a framework integrating a novel contextual bandit algorithm with hierarchical operational space control to learn user contact preferences and translate them into low-level pose and force control policies. PrioriTouch minimizes user discomfort by initially gathering real-world feedback and subsequently refining the policy using simulation-in-the-loop, thus avoiding unsafe user experimentation. Guided by insights from a user study on physical assistance preferences, we rigorously evaluate PrioriTouch in extensive simulation and real-world experiments, demonstrating effective adaptation to user contact preferences, maintained task performance, and enhanced safety and comfort.

Cite this Paper

BibTeX
@InProceedings{pmlr-v305-madan25a, title = {PrioriTouch: Adapting to User Contact Preferences for Whole-Arm Physical Human-Robot Interaction}, author = {Madan, Rishabh and Lin, Jiawei and Goel, Mahika and Li, Amber and Xie, Angchen and Liang, Xiaoyu and Lee, Marcus and Guo, Justin and Thakkar, Pranav N. and Banerjee, Rohan and Barreiros, Jose and Tsui, Kate and Silver, Tom and Bhattacharjee, Tapomayukh}, booktitle = {Proceedings of The 9th Conference on Robot Learning}, pages = {1515--1530}, year = {2025}, editor = {Lim, Joseph and Song, Shuran and Park, Hae-Won}, volume = {305}, series = {Proceedings of Machine Learning Research}, month = {27--30 Sep}, publisher = {PMLR}, pdf = {https://raw.githubusercontent.com/mlresearch/v305/main/assets/madan25a/madan25a.pdf}, url = {https://proceedings.mlr.press/v305/madan25a.html}, abstract = {Many robot caregiving tasks, such as bathing, dressing, and transferring, require a robot arm to make contact with a human body at multiple points rather than solely at the end effector. However, varied human touch preferences can lead to unsafe or uncomfortable multi-contact interactions. To address this, we introduce PrioriTouch, a framework integrating a novel contextual bandit algorithm with hierarchical operational space control to learn user contact preferences and translate them into low-level pose and force control policies. PrioriTouch minimizes user discomfort by initially gathering real-world feedback and subsequently refining the policy using simulation-in-the-loop, thus avoiding unsafe user experimentation. Guided by insights from a user study on physical assistance preferences, we rigorously evaluate PrioriTouch in extensive simulation and real-world experiments, demonstrating effective adaptation to user contact preferences, maintained task performance, and enhanced safety and comfort.} }
Endnote
%0 Conference Paper %T PrioriTouch: Adapting to User Contact Preferences for Whole-Arm Physical Human-Robot Interaction %A Rishabh Madan %A Jiawei Lin %A Mahika Goel %A Amber Li %A Angchen Xie %A Xiaoyu Liang %A Marcus Lee %A Justin Guo %A Pranav N. Thakkar %A Rohan Banerjee %A Jose Barreiros %A Kate Tsui %A Tom Silver %A Tapomayukh Bhattacharjee %B Proceedings of The 9th Conference on Robot Learning %C Proceedings of Machine Learning Research %D 2025 %E Joseph Lim %E Shuran Song %E Hae-Won Park %F pmlr-v305-madan25a %I PMLR %P 1515--1530 %U https://proceedings.mlr.press/v305/madan25a.html %V 305 %X Many robot caregiving tasks, such as bathing, dressing, and transferring, require a robot arm to make contact with a human body at multiple points rather than solely at the end effector. However, varied human touch preferences can lead to unsafe or uncomfortable multi-contact interactions. To address this, we introduce PrioriTouch, a framework integrating a novel contextual bandit algorithm with hierarchical operational space control to learn user contact preferences and translate them into low-level pose and force control policies. PrioriTouch minimizes user discomfort by initially gathering real-world feedback and subsequently refining the policy using simulation-in-the-loop, thus avoiding unsafe user experimentation. Guided by insights from a user study on physical assistance preferences, we rigorously evaluate PrioriTouch in extensive simulation and real-world experiments, demonstrating effective adaptation to user contact preferences, maintained task performance, and enhanced safety and comfort.
APA
Madan, R., Lin, J., Goel, M., Li, A., Xie, A., Liang, X., Lee, M., Guo, J., Thakkar, P.N., Banerjee, R., Barreiros, J., Tsui, K., Silver, T. & Bhattacharjee, T.. (2025). PrioriTouch: Adapting to User Contact Preferences for Whole-Arm Physical Human-Robot Interaction. Proceedings of The 9th Conference on Robot Learning, in Proceedings of Machine Learning Research 305:1515-1530 Available from https://proceedings.mlr.press/v305/madan25a.html.

Related Material