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:20260817T171536Z
LOCATION:Hall B7 (1)\, B Block\, Level 7
DTSTART;TZID=Asia/Tokyo:20241204T144500
DTEND;TZID=Asia/Tokyo:20241204T145600
UID:siggraphasia_SIGGRAPH Asia 2024_sess120_papers_586@linklings.com
SUMMARY:Robot Motion Diffusion Model: Motion Generation for Robotic Charac
 ters
DESCRIPTION:Agon Serifi (ETH Zürich, Disney Research); Ruben Grandia and E
 spen Knoop (Disney Research); Markus Gross (ETH Zürich, Disney Research); 
 and Moritz Bächer (Disney Research)\n\nRecent advancements in generative m
 otion models have achieved remarkable results, enabling the synthesis of l
 ifelike human motions from textual descriptions. These kinematic approache
 s, while visually appealing, often produce motions that fail to adhere to 
 physical constraints, resulting in artifacts that impede real-world deploy
 ment. To address this issue, we introduce a novel method that integrates k
 inematic generative models with physics-based character control. Our appro
 ach begins by training a reward surrogate to predict the performance of th
 e downstream non-differentiable control task, offering an efficient and di
 fferentiable loss function. This reward model is then employed to fine-tun
 e a baseline generative model, ensuring that the generated motions are not
  only diverse but also physically plausible for real-world scenarios. The 
 outcome of our processing is the Robot Motion Diffusion Model (RobotMDM), 
 a text-conditioned kinematic diffusion model that interfaces with a reinfo
 rcement learning-based tracking controller. We demonstrate the effectivene
 ss of this method on a challenging humanoid robot, confirming its practica
 l utility and robustness in dynamic environments.\n\nRegistration Category
 : Full Access, Full Access Supporter\n\nLanguage Format: English Language\
 n\nSession Chair: Hao (Richard) Zhang (Simon Fraser University, Augmenta)\
 n\n
URL:https://asia.siggraph.org/2024/program/?id=papers_586&sess=sess120
END:VEVENT
END:VCALENDAR