SixthSense: Task-Agnostic Proprioception-Only Whole-Body Wrench Estimation for Humanoids

2026-05-02

AI Summary

Humanoid robots often struggle to accurately estimate external wrenches due to a lack of reliable contact perception, which limits their performance in practical tasks. This work proposes a task-agnostic approach that relies solely on proprioceptive and IMU data to jointly estimate the timing, location, and magnitude of whole-body contacts by modeling the spatiotemporal sparsity of contact events through conditional flow matching. The method requires no external sensors or predefined contact assumptions, enabling plug-and-play whole-body wrench estimation for the first time. It significantly enhances the reliability of collision detection, physical human–robot interaction, and force-feedback teleoperation across diverse scenarios, including standing, walking, and full-body motion tracking.

Abstract

Humanoid robots are entering our physical world at scale, yet as oversized toys--good at singing and dancing, but short on force-interaction capabilities for practical tasks. Bridging this gap necessitates prioritizing reliable contact perception as a fundamental requirement. Estimating external wrenches in humanoids is complicated by floating-base dynamics and indeterminate contact locations. Existing analytical frameworks require idealistic assumptions and hard-to-obtain measurements, which are often unavailable in practice. To bridge this gap, we propose SixthSense, a task-agnostic approach that infers whole-body contact timing, location, and wrenches from proprioception and IMU data alone. To capture the multi-modal dynamics between unstructured contact inputs and the uncertain motion outputs, we employ conditional flow matching to tokenize proprioceptive histories and estimate a spatiotemporally sparse contact-event flow. SixthSense serves as a plug-and-play perception module for applications including collision detection, physical human-robot interaction, and force-feedback teleoperation. Experiments across standing, walking, and whole-body motion-tracking policies showcased unprecedented performance in diverse behaviors.

Problem

• wrench estimation
• humanoid robots
• contact perception
• proprioception
• floating-base dynamics

Innovation

• proprioception-only
• whole-body wrench estimation
• conditional flow matching
• task-agnostic perception
• contact-event detection