ReflexArc: Bottom-Up Integration of Instantaneous Tactile Reflexes into Whole-Body Humanoid Control
ReflexArc: Bottom-Up Integration of Instantaneous Tactile Reflexes into Whole-Body Humanoid Control
This paper introduces ARC (Ascending Reflex Control), a framework that integrates instantaneous tactile reflexes with learned whole-body control for humanoid robots. The system detects hazardous contacts via full-body tactile sensing, solves withdrawal poses within 10 ms, and combines reflex-in-the-loop policy learning to maintain balance and complete recovery tasks. Evaluated on a Unitree G1, ARC reduces the first force peak by 34.1% compared to a delayed reflex baseline and lowers contact impulse from 0.2632 N s to 0.1225 N s. The authors also demonstrate protected motion tracking, teleoperation, and vision-free obstacle traversal guided by foot tactile sensing.
The authors present ARC, a control framework integrating instantaneous tactile reflexes with learned whole-body policies to reduce impact forces and improve withdrawal responses on a Unitree G1 humanoid.
Source: arXiv Robotics — research abstracts · Read original article ↗
Loading article text…
What the source reports
Publisher-reported claims, with original evidence. These results have not been independently verified by RoboSignal.
Reported numbers
contact impulse reduction
0.1225
View original evidence
p policy learning, reflex-transparent policy alignment, and reflex-informed policy adaptation to sustain the withdrawal, maintain balance, and complete longer-horizon recovery. We deploy ARC on a Unitree G1 and evaluate three categories of hazardous contact. Compared with an otherwise identical reflex delayed by 20 ms, ARC reduces the first force peak by 34.1% and reduces the contact impulse from 0.2632 N s to 0.1225
Open source S4contact impulse before reduction
0.2632
View original evidence
p policy learning, reflex-transparent policy alignment, and reflex-informed policy adaptation to sustain the withdrawal, maintain balance, and complete longer-horizon recovery. We deploy ARC on a Unitree G1 and evaluate three categories of hazardous contact. Compared with an otherwise identical reflex delayed by 20 ms, ARC reduces the first force peak by 34.1% and reduces the contact impulse from 0.2632 N s to 0.1225
Open source S4reflex pathway latency
10
View original evidence
ottom-up framework that integrates instantaneous tactile reflexes with learned whole-body control. ARC consists of two components: an instantaneous reflex pathway and bottom-up reflex integration. The reflex pathway detects hazardous contacts from full-body tactile sensing, solves a whole-body withdrawal pose, and delivers it directly to the motors within 10 ms. Bottom-up reflex integration combines reflex-in-the-loo
Open source S3policy learning frequency
50
View original evidence
arXiv:2611.04424v1 Announce Type: new Abstract: Humanoid robots operating in cluttered environments must react to unexpected contacts before the first impact peak, which arrives within tens of milliseconds. Learned whole-body policies running at 50 Hz introduce a policy-cycle delay and may suppress a protective withdrawal as a tracking error. To address these challenges, we propose ARC (Ascending Reflex Control), a b
Open source S2reflex delay comparison
20
View original evidence
p policy learning, reflex-transparent policy alignment, and reflex-informed policy adaptation to sustain the withdrawal, maintain balance, and complete longer-horizon recovery. We deploy ARC on a Unitree G1 and evaluate three categories of hazardous contact. Compared with an otherwise identical reflex delayed by 20 ms, ARC reduces the first force peak by 34.1% and reduces the contact impulse from 0.2632 N s to 0.1225
Open source S4
- code: not_reported
- weights: not_reported
- dataset: not_reported
- paper: not_reported
Source excerpts and review record
Automatically extracted; no manual editorial approval recorded.
p policy learning, reflex-transparent policy alignment, and reflex-informed policy adaptation to sustain the withdrawal, maintain balance, and complete longer-horizon recovery. We deploy ARC on a Unitree G1 and evaluate three categories of hazardous contact. Compared with an otherwise identical reflex delayed by 20 ms, ARC reduces the first force peak by 34.1% and reduces the contact impulse from 0.2632 N s to 0.1225
Open source S4
ottom-up framework that integrates instantaneous tactile reflexes with learned whole-body control. ARC consists of two components: an instantaneous reflex pathway and bottom-up reflex integration. The reflex pathway detects hazardous contacts from full-body tactile sensing, solves a whole-body withdrawal pose, and delivers it directly to the motors within 10 ms. Bottom-up reflex integration combines reflex-in-the-loo
Open source S3
arXiv:2611.04424v1 Announce Type: new Abstract: Humanoid robots operating in cluttered environments must react to unexpected contacts before the first impact peak, which arrives within tens of milliseconds. Learned whole-body policies running at 50 Hz introduce a policy-cycle delay and may suppress a protective withdrawal as a tracking error. To address these challenges, we propose ARC (Ascending Reflex Control), a b
Open source S2
N s (a 2.15x ratio). We further demonstrate protected motion tracking, teleoperation, and vision-free obstacle traversal guided by foot tactile sensing. Project page: https://minusmodulo.github.io/reflexarc-website/
Open source S5
Source：arXiv Robotics — research abstracts · arxiv.org