A huge challenge is to simulate tens of thousands of agents in real-time where they pro-actively and realistically avoid collisions with each other and with obstacles present in their environment, especially near narrow passages where cooperative behavior is required. This environment contains semantic information, such as roads or dangerous areas, is usually three-dimensional and can dynamically change.
We study how we can automatically create a data structure that represents the walkable surfaces in virtual environments, and how it can be updated dynamically and efficiently when it changes. We refer to this structure as a navigation mesh. This mesh enables efficient crowd simulation, which is our next topic of research. We study and develop a crowd simulation framework and its components, which ranges from global planning to local animation. We create models for realistic crowd behaviors, which includes studying how people and groups of people move and avoid collisions in such environments, based on agent profiles and semantics.
We continuously integrate our research into our UU Crowd simulation software which is used by many research institutes and companies, including our own startup, uCrowds. We run simulations in realistic environments and game levels to study the effectiveness of our methods.
This way, we contribute to a more safe and fun world.
Video of the week
Real-time, iteractive crowd simulation of 100000 agents
We are proud to present our next performance update: interactive simulation of 100 000 pedestrians.
While our simulation engine could already easily handle 100K pedestrians in real-time, it took some effort to visualize and animate the crowd with an acceptable frame rate (42 fps on a fast PC/GPU) in a 3D-environment. This has been made possible by current and previous students from our Game and Media Technology master program (Mèir, Stijn, Yiran, Nick, Ricky and Anne) and by the powerful Unity3D game engine, in collaboration with our acedemic spinoff, uCrowds.
During these strange Covid-times, crowds of 100K people are rare. However, smaller crowds (of say 20 000 people) can be simulated >32 times real-time. In that case, the simulation of a 4-hour event takes 7.5 minutes. Hence, faster iteration times and better interactivity are now possible.
The lastest news items can be found on our media page.
Click on the pictures for more information about our news items.
World Economic Forum has created a video on our R&D
The video was entitled: Coronavirus using crowd simulation to encourage social distancing.
Keynote at Dutch Benelux Simulation Society
The talk's title was: A computational simulation model of pedestrian navigation in interactive environments. From normal operation to the inclusion of social distancing.
Interactive simulation of 100 000 agents in 3D
While our simulation engine could already easily handle 100K pedestrians in real-time, it took some effort to visualize and animate the crowd with an acceptable frame rate (42 fps on a fast PC/GPU) in a 3D-environment. This has been made possible by the powerful engine of Unity Technologies, and Mèir, Stijn, Yiran, Nick, Ricky and Anne.
Coronavirus: using crowd simulation to encourage social distancing
My article about our R&D on crowd simulation to encourage social distancing was published on theconversation.com. This article was re-published on many sites, including the World economic forum.
Innovation budget of the Ministry of Internal Affairs
Our project 'The city as daily event' has won an Innovation budget of the Dutch Ministry of Internal Affairs. Together with Utrecht, CBS, Movares and uCrowds, we are going to develop a privacy-by-design 'Monitor for crowdedness'.
We celebrate that we have reached the next milestone: real-time simulation of 𝜋 x 100.000 pedestrians on a consumer PC. Thank you Mèir Noordermeer and Stijn Herfst.
Some papers take 7 years to create and publish
After 6 years of coordinating, thinking, researching, writing, coding, and publishing, our paper entitled 'Comparing navigation meshes: theoretical analysis and practical metrics' has arrived in the journal Computer and Graphics.
Real-time simulation of 220K pedestrians
The only way is up….. And we are above 200k real-time pedestrians on our 12-core, consumer PC :-) Our students from the Utrecht University master, Game and Media Technology, have been working on parallelization on three levels. Read more about our metrics and performance simulation.
Spread, spread, spread: how the corona crisis changes our society
The New Scientist has written an article in the series "Voorbij de waan" about Spread, spread, spread: how the corona crisis changes our society, and how crowd simulation R&D of Utrecht University and uCrowds can help.
Software for the 1.5-meter society
I was interviewed about our crowd simulation software for the 1.5-meter society by the Dutch news paper, Algemeen Dagblad. 'Utrechtse informaticus ontwerpt unieke software voor 1,5 metermaatschappij: 'We kunnen van tevoren bepalen hoe druk het ergens wordt''
CLARIN Café I: CLARIN in Times of Corona
I gave a talk in the virtual CLARIN cafe, in which I showcased the relevance of my work on crowd behaviour during social distancing times.
Computer models for safely organising buildings and public space
Utrecht University writes about 'Computer science researcher Roland Geraerts, together with his startup company uCrowds, has adapted his crowd simulation software for 1.5-meter social distancing measures'.
Innovation with a touch
1.5 year ago, we started to develop a user interface from the ground up. We experimented with scanning physical blocks, with scanning a line drawing, and with editing the logic and scanned environment by touch.
Appearance in Dutch TV show, How it's done
The Dutch TV show, How it's done, has aired an item about our crowd simulator R&D on RTL Z.
Interview by Nieuwsuur about social distancing
The Dutch news show, Nieuwsuur, has just aired an item about the influence of social distancing on public transportation. What will become the 'new normal'?
Crowd simulation and social distancing
We celebrate that we have successfully integrated the basis concept of social distancing into our pedestrian simulator, based on research done at Utrecht University and our startup, uCrowds.
Automatic 3D topography model creation
We celebrate that we have added the functionality to create a simple 3D model of an arbitrary area in the Netherlands in our interactive, real-time pedestrian simulator.
I am an assistant professor at the Geometric Computing group in the Department of Information and Computing Sciences at Utrecht University
in the Netherlands. There, I obtained my PhD on sampling-based motion planning techniques. In addition, I studied quality aspects of
paths and roadmaps. My current research focuses on path planning and crowd simulation in games and virtual environments.
I currently teach a course on crowd simulation. I have organized the Creative Game Challenge and I am one of the cofounders of the annual
Motion in Games conference.
We have created a software product (SimCrowds) for efficient crowd simulating in multi-layered 3D dynamic
environments. We use the software for e.g. preparations of the Grand Départ of the Tour de France, and for evacuation studies in the North/South metro line in Amsterdam.
Welcome on my home page on crowd simulation and motion planning