BEGIN:VCALENDAR
VERSION:2.0
PRODID:Linklings LLC
BEGIN:VTIMEZONE
TZID:Asia/Seoul
X-LIC-LOCATION:Asia/Seoul
BEGIN:STANDARD
TZOFFSETFROM:+0900
TZOFFSETTO:+0900
TZNAME:KST
DTSTART:18871231T000000
DTSTART:19881009T020000
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
DTSTAMP:20230103T035307Z
LOCATION:Room 324\, Level 3\, West Wing
DTSTART;TZID=Asia/Seoul:20221206T140000
DTEND;TZID=Asia/Seoul:20221206T153000
UID:siggraphasia_SIGGRAPH Asia 2022_sess154_papers_589@linklings.com
SUMMARY:Transformer Inertial Poser: Real-time Human Motion Reconstruction 
 from Sparse IMUs with Simultaneous Terrain Generation
DESCRIPTION:Technical Communications, Technical Papers\n\nTransformer Iner
 tial Poser: Real-time Human Motion Reconstruction from Sparse IMUs with Si
 multaneous Terrain Generation\n\nJiang, Ye, Gopinath, Won, Winkler...\n\nR
 eal-time human motion reconstruction from a sparse set of (e.g. six) weara
 ble IMUs provides a non-intrusive and economic approach to motion capture.
  Without the ability to acquire position information directly from IMUs, r
 ecent works took data-driven approaches that utilize large human motion da
 tasets to tackle this under-determined problem. Still, challenges remain s
 uch as temporal consistency, drifting of global and joint motions, and div
 erse coverage of motion types on various terrains. We propose a novel meth
 od to simultaneously estimate full-body motion and generate plausible visi
 ted terrain from only six IMU sensors in real-time. Our method incorporate
 s 1. a conditional Transformer decoder model giving consistent predictions
  by explicitly reasoning prediction history, 2. a simple yet general learn
 ing target named "stationary body points” (SBPs) which can be stably predi
 cted by the Transformer model and utilized by analytical routines to corre
 ct joint and global drifting, and 3. an algorithm to generate regularized 
 terrain height maps from noisy SBP predictions which can in turn correct n
 oisy global motion estimation. We evaluate our framework extensively on sy
 nthesized and real IMU data, and with real-time live demos, and show super
 ior performance over strong baseline methods.\n\nRegistration Category: FU
 LL ACCESS, ON-DEMAND ACCESS\n\nLanguage: ENGLISH\n\nFormat: IN-PERSON, ON-
 DEMAND
URL:https://sa2022.siggraph.org/en/full-program/?id=papers_589&sess=sess15
 4
END:VEVENT
END:VCALENDAR