A Microscopic Pedestrian Crowd Simulation System
PEDSIM is a microscopic pedestrian crowd simulation system. It is suitable for use in crowd simulations (e.g. indoor evacuation simulation, large scale outdoor simulations), where one is interested in output like pedestrian density or evacuation time. Also, the quality of the individual agent's trajectory is high enough for creating massive pedestrian crowd animations (e.g. for motion pictures or architectural visualization). Since libpedsim is easy to use and extend, it is a good starting point for science projects. See the examples page for example pictures, short movies, and for screenshots.
The PEDSIM library allows you to use pedestrian dynamics in your own software. Based on pure C++ without additional packages, it runs virtually on every operating system. The PEDSIM Demo Application gives you a quick overview of the capabilities, and is a starting point for your own experiments. It is built using the Qt Framework. You can access precompiled binary packages and the source code at the download page.
The pedestrians are visible on the user interface in real-time. Using the file or tcp/stream-based output, batch or real-time processing is possible. Typically, the output from PEDSIM is fed into a rendering engine, where realistically looking humans are created. These humans walk based on the trajectories generated by PEDSIM.
While the used simulation model is simple and easy to understand, it is possible to extend the agent's behavior. Several such mental strategies are included, and adding your own code is possible.
If you have questions regarding PEDSIM or pedestrian simulations in general, don't hesitate to ask. Further development efforts depend on feedback. Please submit bugs, patches, critism, ideas, or directions in which development should go.
|2015-12-10 :|| A fresh version of PEDSIM is available (Latest Version, Repo Snapshot). It was created automatically from the
source code repository.
Update Comment: Fixed memory leak in example01.
Version 2.3 has been released!
The features have been available in the repo snapshot for two years, but finally a consistent and stable release is available. Version 2.3 contains the latest improvements and features. Mostly small fixes accumulated over time, but notably one big patch provided by Sven Wehner. The model is now in line again with the publications by Prof. Helbing and his team.