· Contributors · Organizations · Search
CCP: Configurable Crowd Profiles
Research & Education
Full Conference Supporter
Exhibitor Additional Full Conference
Exhibitor Full Conference
This session WILL NOT be recorded.
LocationEast Building, Room 1-3
DescriptionWe present a reinforcement learning based framework to learn multiple agent behaviors concurrently. We optimize the agent policy by varying the importance of selected behaviors (goal seeking, collision avoidance, interaction with environment, and grouping) while training. These combinations define multiple different profiles that can dynamically be defined during simulation time.