Natural Behavior Planning for Autonomous Agents in Dense Environments

1
Natural Behavior Planning for Autonomous Agents
in Dense Environments

• COMP790-58 Course Project Proposal
• Tao Yu

2
Crowd simulation

• Global path planning
• Local collision avoidance
• Human behavior simulation

3
Behavior model

• High level
• What kind of action to take under current
  situation
• Collision response -gt interaction
• Low level
• How to perform the action while satisfying
  constraints (via motion controller or Existing
  motion data)

4
Previous work

• Behavior planning
• Behavior planning for character animation. SCA05
• Precomputed search trees Planning for
  interactive goal-driven animation. SCA06
• Data-driven models
• Crowds by Example. EG07
• Group Behavior from Video A Data-driven Approach
  to Crowd Simulation. SCA07

5
Previous work (continued)

• HiDAC
• Controlling Individual Agents in High-Density
  Crowd Simulation. SCA07
• Motion graph
• Fast and Accurate Goal-Directed Motion Synthesis
  for Crowds. SCA 05

6
Goals

• Agent motion with visually appealing quality in
  dense environments
• Incorporate agent interaction
• Runtime efficiency