HITTER: A HumanoId Table TEnnis Robot via Hierarchical Planning and Learning

์ €์ž: Zhi Su, Bike Zhang, Nima Rahmanian, Yuman Gao, Qiayuan Liao, Caitlin Regan, Koushil Sreenath, S. Shankar Sastry | ๋‚ ์งœ: 2025-09-04 | DOI: 10.48550/arXiv.2508.21043 📄 PDF


Essence

Figure 2

Fig. 2: System overview. (a) The racket is mounted on the robotโ€™s right wrist using a 3D-printed connector, and the ball

ํœด๋จธ๋…ธ์ด๋“œ ๋กœ๋ด‡์ด ํƒ๊ตฌ๋ฅผ ํ•˜๊ธฐ ์œ„ํ•œ ๊ณ„์ธต์  ํ”„๋ ˆ์ž„์›Œํฌ๋ฅผ ์ œ์‹œํ•˜๋ฉฐ, model-based planner์™€ RL ๊ธฐ๋ฐ˜ whole-body controller๋ฅผ ํ†ตํ•ฉํ•˜์—ฌ sub-second ๋ฐ˜์‘ ์‹œ๊ฐ„ ๋‚ด์— ์ดˆ๋‹น 5 m/s ์ด์ƒ์˜ ๋ณผ์„ ์ฒ˜๋ฆฌํ•œ๋‹ค.

Motivation

Achievement

Figure 1

Fig. 1: Humanoid table tennis rallies. Our system enables both humanoid-humanoid (left) and humanoid-human (right) match

How

Figure 2

Fig. 2: System overview. (a) The racket is mounted on the robotโ€™s right wrist using a 3D-printed connector, and the ball

Originality

Limitation & Further Study

Evaluation

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

์ดํ‰: ๋ณธ ๋…ผ๋ฌธ์€ humanoid table tennis๋ฅผ ํ†ตํ•ด ๊ณ ์† ๋™์  ํ™˜๊ฒฝ์—์„œ์˜ ์ „์‹  ์ œ์–ด ๋ฐ ์ƒํ˜ธ์ž‘์šฉ์„ ์ฒ˜์Œ์œผ๋กœ ์„ฑ๊ณต์ ์œผ๋กœ ์‹œ์—ฐํ•˜์˜€์œผ๋ฉฐ, ๊ณ„์ธต์  planning-control ํ†ตํ•ฉ๊ณผ minimal human references๋ฅผ ํ†ตํ•œ ์šฐ์•„ํ•œ ์ ‘๊ทผ๋ฒ•์ด ์ธ์ƒ์ ์ด๋‹ค. ์‹ค์ œ ์„ธ๊ณ„ ๊ฒ€์ฆ(106 ์—ฐ์† ์ƒท)์€ ๋ฐฉ๋ฒ•๋ก ์˜ ์‹ค์šฉ์„ฑ์„ ๊ฐ•๋ ฅํžˆ ์ž…์ฆํ•œ๋‹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˜์ •