GentleHumanoid: Learning Upper-body Compliance for Contact-rich Human and Object Interaction

์ €์ž: Qingzhou Lu, Yao Feng, Baiyu Shi, Michael Piseno, Zhenan Bao, C. Karen Liu | ๋‚ ์งœ: 2025-11-06 | DOI: 10.48550/arXiv.2511.04679 📄 PDF


Essence

Figure 1

Fig. 1: GentleHumanoid learns a universal whole-body control policy with upper-body compliance and tunable force limits.

GentleHumanoid๋Š” impedance control์„ whole-body motion tracking ์ •์ฑ…์— ํ†ตํ•ฉํ•˜์—ฌ humanoid ๋กœ๋ด‡์˜ ์ƒ์ฒด compliance๋ฅผ ํ•™์Šตํ•˜๋Š” ํ”„๋ ˆ์ž„์›Œํฌ์ด๋‹ค. ์ด๋Š” human motion data์—์„œ ์ƒ˜ํ”Œ๋งํ•œ spring-based formulation์„ ํ†ตํ•ด resistive contact์™€ guiding contact๋ฅผ ํ†ต์ผ์ ์œผ๋กœ ๋ชจ๋ธ๋งํ•œ๋‹ค.

Motivation

Achievement

Figure 1

Fig. 1: GentleHumanoid learns a universal whole-body control policy with upper-body compliance and tunable force limits.

How

Figure 2

Fig. 2: Overview framework. (a) Reference dynamics: impedance-based dynamics integrate driving forces (for motion

Originality

Limitation & Further Study

Evaluation

Novelty: 4/5 Technical Soundness: 3/5 Significance: 4/5 Clarity: 4/5 Overall: 4/5

์ดํ‰: GentleHumanoid๋Š” humanoid ๋กœ๋ด‡์˜ ์•ˆ์ „ํ•œ human-robot physical interaction์„ ์œ„ํ•œ ์‹ค์งˆ์ ์ด๊ณ  ์ฐฝ์˜์ ์ธ ์†”๋ฃจ์…˜์„ ์ œ์‹œํ•œ๋‹ค. Unified spring-based formulation๊ณผ human motion data ๊ธฐ๋ฐ˜ contact modeling์˜ ์กฐํ•ฉ์€ novelํ•˜๋ฉฐ, ์‹ค์ œ Unitree G1์—์„œ์˜ ๊ฒ€์ฆ๊ณผ custom pressure-sensing ํ‰๊ฐ€ ๋ฐฉ๋ฒ•๋ก ์€ ๋…ผ๋ฌธ์˜ ์‹ ๋ขฐ์„ฑ์„ ๋†’์ธ๋‹ค.

← ๋ชฉ๋ก์œผ๋กœ ๋Œ์•„๊ฐ€๊ธฐ

๐ŸŽง Audio Overview

์ด ๋…ผ๋ฌธ ๋ฆฌ๋ทฐ๋ฅผ ํŒŸ์บ์ŠคํŠธํ˜• ์˜ค๋””์˜ค๋กœ ์ƒ์„ฑํ•ฉ๋‹ˆ๋‹ค. (Gemini ยท ํ‚ค๋Š” ๋ธŒ๋ผ์šฐ์ €์—๋งŒ ์ €์žฅ ยท ์™„์„ฑ๋ณธ์€ ์ด๋ฉ”์ผ๋กœ๋„ ์ „์†ก)
โ–ธ ๊ณ ๊ธ‰: ๊ตฌ์„ฑ ๋ฐฉํ–ฅ(๋Œ€๋ณธ ์ž‘์„ฑ ์ง€์นจ) ์ง์ ‘ ์ˆ˜์ •