BEGIN:VCALENDAR VERSION:2.0 PRODID:IEEE vTools.Events//EN CALSCALE:GREGORIAN BEGIN:VTIMEZONE TZID:America/New_York BEGIN:DAYLIGHT DTSTART:20260308T030000 TZOFFSETFROM:-0500 TZOFFSETTO:-0400 RRULE:FREQ=YEARLY;BYDAY=2SU;BYMONTH=3 TZNAME:EDT END:DAYLIGHT BEGIN:STANDARD DTSTART:20261101T010000 TZOFFSETFROM:-0400 TZOFFSETTO:-0500 RRULE:FREQ=YEARLY;BYDAY=1SU;BYMONTH=11 TZNAME:EST END:STANDARD END:VTIMEZONE BEGIN:VEVENT DTSTAMP:20260602T233245Z UID:459E3596-3F9B-4BB6-83C1-3F249BBF01FB DTSTART;TZID=America/New_York:20260602T100000 DTEND;TZID=America/New_York:20260602T120000 DESCRIPTION:Autonomous mobility systems increasingly rely on collaborative perception to overcome occlusion\, limited field of view\, and social navi gation challenges in dynamic environments. However\, effective collaborati on is not simply about sharing more sensing data\; it requires identifying information that is semantically valuable for a mobility agent’s task\, decision-making\, and evolving situational awareness. This talk explores how collaborative perception can move from extensive sensing to comprehens ive understanding through world models. We begin with recent advances in v ision-language models for semantic-aware perception\, while highlighting k ey limitations: insufficient sensing data for reliable reasoning and the t ime-varying nature of perception evidence. To address these challenges\, w e introduce world models for evaluating collaboration policies that mainta in reliable situational awareness as sensing coverage\, mobility patterns\ , and communication conditions evolve. By predicting whether and how colla boration can improve semantic confidence under evolving sensing\, mobility \, and communication conditions\, this approach transforms collaboration f rom reactive raw-data sharing into predictive\, semantic-aware communicati on and policy reasoning\, enabling autonomous systems to proactively ident ify efficient collaboration patterns.\n\nSpeaker(s): Mushu Li\n\nRoom: 415 2\, Bldg: Centre for Environmental & Information Technology (EIT)\, Univer sity of Waterloo\, Waterloo\, Ontario\, Canada LOCATION:Room: 4152\, Bldg: Centre for Environmental & Information Technolo gy (EIT)\, University of Waterloo\, Waterloo\, Ontario\, Canada ORGANIZER:leil@uoguelph.ca SEQUENCE:24 SUMMARY:From Sensing to Understanding: World Models for Semantic-Aware Coll aborative Perception URL;VALUE=URI:https://events.vtools.ieee.org/m/560981 X-ALT-DESC:Description: <br /><p class="MsoNormal">Autonomous mobility syst ems increasingly rely on collaborative perception to overcome occlusion\, limited field of view\, and social navigation challenges in dynamic enviro nments. However\, effective collaboration is not simply about sharing more sensing data\; it requires identifying information that is semantically v aluable for a mobility agent&rsquo\;s task\, decision-making\, and evolvin g situational awareness. This talk explores how collaborative perception c an move from extensive sensing to comprehensive understanding through worl d models. We begin with recent advances in vision-language models for sema ntic-aware perception\, while highlighting key limitations: insufficient s ensing data for reliable reasoning and the time-varying nature of percepti on evidence. To address these challenges\, we introduce world models for e valuating collaboration policies that maintain reliable situational awaren ess as sensing coverage\, mobility patterns\, and communication conditions evolve. By predicting whether and how collaboration can improve semantic confidence under evolving sensing\, mobility\, and communication condition s\, this approach transforms collaboration from reactive raw-data sharing into predictive\, semantic-aware communication and policy reasoning\, enab ling autonomous systems to proactively identify efficient collaboration pa tterns.</p> END:VEVENT END:VCALENDAR