BEGIN:VCALENDAR
VERSION:2.0
PRODID:Linklings LLC
BEGIN:VTIMEZONE
TZID:Asia/Tokyo
X-LIC-LOCATION:Asia/Tokyo
BEGIN:STANDARD
TZOFFSETFROM:+0900
TZOFFSETTO:+0900
TZNAME:JST
DTSTART:18871231T000000
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
DTSTAMP:20260817T171533Z
LOCATION:Hall B7 (1)\, B Block\, Level 7
DTSTART;TZID=Asia/Tokyo:20241206T145600
DTEND;TZID=Asia/Tokyo:20241206T150800
UID:siggraphasia_SIGGRAPH Asia 2024_sess150_papers_1155@linklings.com
SUMMARY:Synchronize Dual Hands for Physics-Based Dexterous Guitar Playing
DESCRIPTION:Pei Xu and Ruocheng Wang (Stanford University)\n\nWe present a
  novel approach to synthesize dexterous motions for physically simulated h
 ands in tasks that require coordination between the control of two hands w
 ith high temporal precision. Instead of directly learning a joint policy t
 o control two hands, our approach performs bimanual control through cooper
 ative learning where each hand is treated as an individual agent. The indi
 vidual policies for each hand are first trained separately, and then synch
 ronized through latent space manipulation in a centralized environment to 
 serve as a joint policy for two-hand control. By doing so, we avoid direct
 ly performing policy learning in the joint state-action space of two hands
  with higher dimensions, greatly improving the overall training efficiency
 . We demonstrate the effectiveness of our proposed approach in the challen
 ging guitar-playing task. The virtual guitarist trained by our approach ca
 n synthesize motions from unstructured reference data of general guitar-pl
 aying practice motions, and accurately play diverse rhythms with complex c
 hord pressing and string picking patterns based on the input guitar tabs t
 hat do not exist in the references. Along with this paper, we provide the 
 motion capture data that we collected as the reference for policy training
 .\n\nRegistration Category: Full Access, Full Access Supporter\n\nLanguage
  Format: English Language\n\nSession Chair: Li-Yi Wei (Adobe Research)\n\n
URL:https://asia.siggraph.org/2024/program/?id=papers_1155&sess=sess150
END:VEVENT
END:VCALENDAR