Crowd Simulation (CSIM)

Exploring GPU-accelerated crowd simulation using Position-Based Dynamics (PBD) and a novel density-driven Level-of-Detail (LOD) system - 2026

Project Overview

Crowd Simulation (CSIM) is a project focused on developing a GPU-accelerated crowd simulation system using Position-Based Dynamics (PBD) and a novel density-driven Level-of-Detail (LOD) system. I explored various techniques to optimize the simulation of large crowds, including spatial partitioning and parallel processing on the GPU. The project aimed to explore different approaches, including force-based method and position-based method, to create a scalable and efficient crowd simulation system.

Check the repo here

My Contributions

Programming

  • Built a force-based solver.
  • Built multithreaded CPU versions of both the force-based and position-based solvers.
  • Built a GPU-accelerated versions of both solvers using compute shaders for large-scale, real-time crowd simulation.
CSIM - Project Report