BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:PRECOG: PREdiction Conditioned On Goals in Visual Multi-Agent Sett
 ings - Rowan McAlister\, University of California\, Berkeley
DTSTART:20190625T123000Z
DTEND:20190625T133000Z
UID:TALK126955@talks.cam.ac.uk
CONTACT:Microsoft Research Cambridge Talks Admins
DESCRIPTION:For autonomous vehicles (AVs) to behave appropriately on roads
  populated by human-driven vehicles\, they must be able to reason about th
 e uncertain intentions and decisions of other drivers from rich perceptual
  information. Towards these capabilities\, we present a probabilistic fore
 casting model of future interactions of multiple agents. We perform both s
 tandard forecasting and conditional forecasting with respect to the AV’s
  goals. Conditional forecasting reasons about how all agents will likely r
 espond to specific decisions of a controlled agent. We train our model on 
 real and simulated data to forecast vehicle trajectories given past positi
 ons and LIDAR. Our evaluation shows that our model is substantially more a
 ccurate in multi-agent driving scenarios compared to existing state-of-the
 -art. Beyond its general ability to perform conditional forecasting querie
 s\, we show that our model’s predictions of all agents improve when cond
 itioned on knowledge of the AV’s intentions\, further illustrating its c
 apability to model agent interactions.
LOCATION:Small Lecture Theatre\, Microsoft Research Ltd\, 21 Station Road\
 , Cambridge\, CB1 2FB
END:VEVENT
END:VCALENDAR