BEGIN:VCALENDAR

VERSION:2.0

PRODID:-//wp-events-plugin.com//6.6.4.4//EN

TZID:Asia/Jerusalem

X-WR-TIMEZONE:Asia/Jerusalem
BEGIN:VEVENT

UID:0-234@aerospace.technion.ac.il

DTSTART;TZID=Asia/Jerusalem:20210623T163000

DTEND;TZID=Asia/Jerusalem:20210623T173000

DTSTAMP:20230409T143701Z

URL:https://aerospace.technion.ac.il/events/merchav-prize-seminar-epistemi
 c-uncertainty-aware-semantic-localization-and-mapping-for-inference-and-be
 lief-space-planning/

SUMMARY:Merchav Prize Seminar | Epistemic Uncertainty Aware Semantic Locali
 zation and Mapping for Inference and Belief Space Planning
DESCRIPTION:Lecturer:Vladimir Tchuiev\n Faculty:Department of Aerospace Eng
 ineering\n Institute:Technion – Israel Institute of Technology\n Locatio
 n:Classroom 165\, ground floor\, Library\, Aerospace Eng.\n Zoom: \n Abstr
 act: \n Details: \n We investigate the problem of autonomous object classi
 fication and semantic SLAM\, which in general exhibits a tight coupling be
 tween classification\, metric SLAM and planning under uncertainty. We cont
 ribute a unified framework for inference and belief space planning (BSP) t
 hat addresses prominent sources of uncertainty in this context: classifica
 tion aliasing (classier cannot distinguish between candidate classes from 
 certain viewpoints)\, classifier epistemic uncertainty (classifier receive
 s data "far" from its training set)\, and localization uncertainty (camera
  and object poses are uncertain). Specifically\, we develop two methods fo
 r maintaining a joint distribution over robot and object poses\, and over 
 posterior class probability vector that considers epistemic uncertainty in
  a Bayesian fashion. The first approach is Multi-Hybrid (MH)\, where multi
 ple hybrid beliefs over poses and classes are maintained to approximate th
 e joint belief over poses and posterior class probability. The second appr
 oach is Joint Lambda Pose (JLP)\, where the joint belief is maintained dir
 ectly using a novel JLP factor. Furthermore\, we extend both methods to BS
 P\, planning while reasoning about future posterior epistemic uncertainty 
 indirectly\, or directly via a novel information-theoretic reward function
 . Both inference methods utilize a novel viewpoint-dependent classifier un
 certainty model that leverages the coupling between poses and classificati
 on scores and predicts the epistemic uncertainty from certain viewpoints. 
 In addition\, this model is used to generate predicted measurements during
  planning. To the best of our knowledge\, this is the first work that reas
 ons about classifier epistemic uncertainty within semantic SLAM and BSP.
CATEGORIES:Seminars
LOCATION:Classroom 165\, ground floor\, Library\, Aerospace Eng.

END:VEVENT

BEGIN:VTIMEZONE

TZID:Asia/Jerusalem

X-LIC-LOCATION:Asia/Jerusalem

BEGIN:DAYLIGHT

DTSTART:20210326T030000

TZOFFSETFROM:+0200

TZOFFSETTO:+0300

TZNAME:IDT

END:DAYLIGHT

END:VTIMEZONE
END:VCALENDAR