BEGIN:VCALENDAR
VERSION:2.0
PRODID:Linklings LLC
BEGIN:VTIMEZONE
TZID:Australia/Melbourne
X-LIC-LOCATION:Australia/Melbourne
BEGIN:DAYLIGHT
TZOFFSETFROM:+1000
TZOFFSETTO:+1100
TZNAME:AEDT
DTSTART:19721003T020000
RRULE:FREQ=YEARLY;BYMONTH=4;BYDAY=1SU
END:DAYLIGHT
BEGIN:STANDARD
DTSTART:19721003T020000
TZOFFSETFROM:+1100
TZOFFSETTO:+1000
TZNAME:AEST
RRULE:FREQ=YEARLY;BYMONTH=10;BYDAY=1SU
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
DTSTAMP:20240214T070250Z
LOCATION:Meeting Room C4.11\, Level 4 (Convention Centre)
DTSTART;TZID=Australia/Melbourne:20231215T161000
DTEND;TZID=Australia/Melbourne:20231215T162000
UID:siggraphasia_SIGGRAPH Asia 2023_sess139_papers_523@linklings.com
SUMMARY:Adaptive Tracking of a Single-Rigid-Body Character in Various Envi
 ronments
DESCRIPTION:Technical Papers\n\nTaesoo Kwon, Taehong Gu, Jaewon Ahn, and Y
 oonsang Lee (Hanyang University)\n\nSince the introduction of DeepMimic [P
 eng et al. 2018], subsequent research\nhas focused on expanding the repert
 oire of simulated motions across various\nscenarios. In this study, we pro
 pose an alternative approach for this goal,\na deep reinforcement learning
  method based on the simulation of a single-\nrigid-body character. Using 
 the centroidal dynamics model (CDM) to express\nthe full-body character as
  a single rigid body (SRB) and training a policy to\ntrack a reference mot
 ion, we can obtain a policy that is capable of adapting\nto various unobse
 rved environmental changes and controller transitions\nwithout requiring a
 ny additional learning. Due to the reduced dimension\nof state and action 
 space, the learning process is sample-efficient. The final\nfull-body moti
 on is kinematically generated in a physically plausible way,\nbased on the
  state of the simulated SRB character. The SRB simulation is\nformulated a
 s a quadratic programming (QP) problem, and the policy outputs\nan action 
 that allows the SRB character to follow the reference motion. We\ndemonstr
 ate that our policy, efficiently trained within 30 minutes on an\nultrapor
 table laptop, has the ability to cope with environments that have\nnot bee
 n experienced during learning, such as running on uneven terrain\nor pushi
 ng a box, and transitions between learned policies, without any\nadditiona
 l learning.\n\nRegistration Category: Full Access\n\nSession Chair: Yuting
  Ye (Reality Labs Research, Meta)
URL:https://asia.siggraph.org/2023/full-program?id=papers_523&sess=sess139
END:VEVENT
END:VCALENDAR