DexCap: Scalable and Portable Mocap Data Collection System for Dexterous Manipulation

์ €์ž: Chen Wang, Haochen Shi, Weizhuo Wang, Ruohan Zhang, Li Fei-Fei, C. Karen Liu | ๋‚ ์งœ: 2024-03-12 | URL: https://arxiv.org/abs/2403.07788 📄 PDF


Essence

Figure 1

Fig. 1: DEXCAP facilitates the in-the-wild collection of high-quality human hand motion capture data and 3D observations

DexCap์€ SLAM๊ณผ ์ „์ž๊ธฐ์žฅ์„ ํ™œ์šฉํ•œ ํœด๋Œ€์šฉ ์† ๋ชจ์…˜์บก์ฒ˜ ์‹œ์Šคํ…œ์ด๋ฉฐ, DexIL์€ ์ด ๋ฐ์ดํ„ฐ๋กœ๋ถ€ํ„ฐ ์—ญ์šด๋™ํ•™๊ณผ point cloud ๊ธฐ๋ฐ˜ ๋ชจ๋ฐฉํ•™์Šต์„ ํ†ตํ•ด ๋กœ๋ด‡์ด ์†๊ฐ€๋ฝ ์กฐ์ž‘์„ ์ง์ ‘ ํ•™์Šตํ•˜๋„๋ก ํ•˜๋Š” ์•Œ๊ณ ๋ฆฌ์ฆ˜์ด๋‹ค.

Motivation

Achievement

Figure 2

Fig. 2: Details of the human system. (a) Our setup includes a 3D-printed rack on a chest harness, featuring a Realsense

How

Figure 4

Fig. 4: Algorithm overview. (a) DEXIL first retargets the DEXCAP data to the robot embodiment by first constructing 3D

Originality

Limitation & Further Study

Evaluation

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

์ดํ‰: DexCap๊ณผ DexIL์€ ํœด๋Œ€์šฉ mocap ์‹œ์Šคํ…œ๊ณผ embodiment gap์„ ๊ทน๋ณตํ•˜๋Š” imitation learning์„ ์ฒ˜์Œ์œผ๋กœ ํ†ตํ•ฉํ•˜์—ฌ in-the-wild ํ™˜๊ฒฝ์—์„œ ๋กœ๋ด‡ ์†๊ฐ€๋ฝ ์กฐ์ž‘ ํ•™์Šต์„ ๊ฐ€๋Šฅํ•˜๊ฒŒ ํ•œ ์šฐ์ˆ˜ํ•œ ๊ธฐ์—ฌ์ด๋ฉฐ, 6๊ฐ€์ง€ ์กฐ์ž‘ ์ž‘์—…์—์„œ ์ผ๊ด€๋œ ์„ฑ๊ณผ๋ฅผ ๋ณด์—ฌ์ค€๋‹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˜์ •