Reproduce “Tactile-Proprioceptive Sensor Fusion for Contact Wrench Estimation in Whole-Body Physical Human-Robot Interaction” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Junha Min, Junghyeon Ma, Jiwung Kwon, Sunggyu Bae, Joohyung Kim, Kyungseo Park
Researchers from DGIST and the University of Illinois Urbana-Champaign developed a hybrid tactile-proprioceptive sensor fusion framework for whole-body physical human-robot interaction. It combines pneumatic robot skins with motor-current-based proprioception, augmented by a Temporal Convolutional Network (TCN) to compensate for complex friction dynamics, reducing the proprioceptive 'dead band' by 88.6% and enabling more responsive kinesthetic teaching.
Key insights
- Tactile sensors can serve as reliable contact indicators, enabling explicit disambiguation of static friction from true external forces in proprioceptive measurements.
- A data-driven TCN can effectively model and compensate for the complex, trajectory-dependent hysteresis of motor friction, especially during static-to-kinetic transitions, which traditional models struggle with.
- Fusing precise static friction estimates from tactile-driven FSM with the TCN's learning enables a unified, low-threshold dead band across various robot interaction states, ensuring seamless transitions.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.