BEGIN:VCALENDAR
VERSION:2.0
PRODID:Linklings LLC
BEGIN:VTIMEZONE
TZID:Asia/Hong_Kong
X-LIC-LOCATION:Asia/Hong_Kong
BEGIN:STANDARD
TZOFFSETFROM:+0800
TZOFFSETTO:+0800
TZNAME:HKT
DTSTART:19911015T033000
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
DTSTAMP:20251218T030656Z
LOCATION:Meeting Room S221\, Level 2
DTSTART;TZID=Asia/Hong_Kong:20251215T171300
DTEND;TZID=Asia/Hong_Kong:20251215T172400
UID:siggraphasia_SIGGRAPH Asia 2025_sess115_papers_2360@linklings.com
SUMMARY:Fast & Stable Control of Coupled Solid-Fluid Dynamic Systems
DESCRIPTION:Jie Chen (VCIP, College of Computer Science, Nankai University
 ); Zherong Pan (LIGHTSPEED); and Bo Ren (VCIP, College of Computer Science
 , Nankai University)\n\nWe propose a Reinforcement Learning (RL) algorithm
  that combines several novel techniques to achieve more stable and robust 
 control results for coupled solid-fluid systems. Our method utilizes the t
 win-delayed actor-critic algorithm to efficiently utilize off-policy data 
 and achieve faster convergence. For more accurate estimations of the value
  function to guide the search of optimal policies, we use the Boltzmann so
 ftmax operator to reduce the bias of estimation. We further introduce a no
 vel two-step Q-value estimator to reduce the well-known under-estimation i
 ssue. Furthermore, to mitigate the requirement of excessive exploration un
 der sparse rewards, we propose the Fluid Effective Domain Guidance (FEDG) 
 algorithm to guide policy exploration, where the policy for an easier task
  is trained jointly with that for a harder task. Put together, our framewo
 rk achieves state-of-the-art performance in complex fluid-solid coupling c
 ontrol benchmarks, delivering stable and reliable performance in both 2D a
 nd 3D tasks over long horizons.\n\nRegistration Category: Full Access, Ful
 l Access Supporter\n\nSession Chair: Bo Ren (TMCC, College of Computer Sci
 ence, Nankai University)\n\n
END:VEVENT
END:VCALENDAR