Behavioral Animation for Crowd Simulation


Funda Durupinar
Ph.D Student
Computer Engineering Department
Bilkent University

Autonomous agents appear in many areas of computer animation, from the simulation of biological systems to computer games. Behavioral animation systems model and simulate autonomous agents that act according to some predefined rules and environmental influences. Behavioral animation covers a wide range of problems that are common with robotics, artificial intelligence and artificial life. However, the main application area of behavioral animation is the situations involving a large number of entities, i.e., crowd simulation. Thus, the tedious and labor-intensive work of motion design process is automated, leaving merely the behavior design to the animator. The main purpose of this thesis study is to develop a behavioral animation system for crowd simulation. The system will include components like perception, memory, learning, motor control, etc. Existing systems in the literature are specific to certain situations only. We aim to develop a generic platform that enables the animator to author certain scenarios with a large number of autonomous agents that perform given tasks. Another issue about the existing systems is their time complexity. They are computationally inefficient (O(n2)) as they require the comparison of each agent with every other agent. We also plan to develop algorithms to decrease the computational cost.


DATE: 26 April, 2007, Thursday@ 10:00