We present a new local collision avoidance algorithm between multiple agents for real-time simulations. Our approach extends the notion of velocity obstacles from robotics and formulates the conditions for collision free navigation as a quadratic optimization problem. We use a discrete optimization method to efficiently compute the motion of each agent. This resulting algorithm can be parallelized by exploiting data-parallelism and thread-level parallelism. The overall approach, ClearPath, is

more »

... neral and can robustly handle dense scenarios with tens or hundreds of thousands of heterogeneous agents in a few milli-seconds. As compared to prior collision avoidance algorithms, we observe more than an order of magnitude performance improvement. † Contact: sjguy@cs.unc.edu Figure 1: Building evacuation: 1, 000 independent agents in different rooms of a building move towards the two exit signs and cause congestion. Our new parallel collision avoidance algorithm, P-ClearPath can efficiently perform local collision avoidance for all agents in such tight packed simulations at 550 FPS on Intel quad-core Xeon (3.14 GHz) processor, and 4, 500 FPS on a 32-core Larrabee processor. Our algorithm is an order of magnitude faster than prior velocity-obstacle based algorithms. Furthermore, applications such as large-scale urban simulations often need to simulate tens or hundreds of thousands of heterogeneous agents at interactive rates. One of our goals in studying the computational issues involved in enabling real-time agent-based simulation is c The Eurographics Association 2009. Stephen J. Guy et al. / ClearPath