Inria Associated Team SIMS

 

Presentation

 
The Inria Associated team SIMS gathers 4 teams working in the domain of virtual human with two specific focus given to motion planning and crowd simulation. Our general objective is to increase the level of realism of human simulation with respect to computing performance.

More precisely, the activity of SIMS is organized along three main axis of research: i) acquisition of motion data,  ii) local models of interaction for crowd simulation, and iii) assimilation of data by simulation models.

Human motion data can be of three kinds. First, empirical data capture the motion of real crowds. They serve as example or ground-truth data for evaluation and parameter setting purposes. Laboratory data are recorded for subjects moving in controlled conditions. Specific factors can be isolated and studied, accurate motion capture can be performed in laboratory conditions. Finally, VR-data capture a real subject moving among a crowd of virtual crowd by using immersive display systems. VR facilitates experiments by requiring only one human subject and by allowing accurate control and reproducibility of experimental conditions. Those three types of data are complementary, and are used for different purposes, they are each 5 of high value for the crowd simulation community.

SIMS wants to develop experiments as well as motion analysis framework to better understand local interactions in crowds. Interactions are characterized by the influence each one’s motion has on neighbors, resulting on correlated motion signal variations. There were very few attempts to detect those correlations in both space and time in order to better understand the nature of these influences (what kind of motion adaptations result from interaction) as well as their extent among neighbors (who is interacting with who among the crowd members). This objective is based on the analysis of empirical data. This analysis should enable us formulating study hypothesis to design both laboratory and VR experiments on crowds.

The SIMS members have pioneered the development of a new generation of crowd simulation algorithms for which each agent predicts other agents motion over a short future time window and adapt their own motion accordingly (e.g., to avoid collision). This principle is more representative of real humans motion because it captures anticipation behavior and enabled crowd simulators to achieve more realistic results.

Partners

 
To meet this objective, SIMS involves 3 teams in the United States:

1.            The GAMMA Research Group, University of North Carolina in Chapel Hill, USA  (gamma.web.unc.edu
2.            The VenLAB, Brown University, Providence, USA (www.cog.brown.edu/research/ven_lab)
3.            The Applied Motion lab, University of Minnesota, Minneapolis, USA (motion.cs.umn.edu

 

Events


March 2017: SIMS organizes the IEEE VR 2017 workshop on Virtual Humans and Crowds for Interactive Environments. Submission deadline is February 3. Get more information on: https://sites.google.com/site/vhcieieeevr2017/

December 2016: Our joint work with UNC is presented at SIGGRAPH Asia 2016 by David Wolinski [Wolinski 2016]. Check the paper for simulation of plane evacuations. 

wolinski 2016

March 2016: We have organized massive experiments on crowd behaviors in Rennes, jointly with W. Warren at Brown University (see pictures below)

     

March 2016: workshop on Virtual Humans and Crowds for Interactive Environments at IEEE VR 2016 in Greenville, USA. See the program and more details on: https://sites.google.com/site/vhcie2016/

March 2016: yearly meeting in Arles, during IEEE VR.

February 2016: Our joint work with UNC will be presented at I3D 2016 by Shonyon Park [Park 2015].

park

June 2015: We organized a transdisciplinary workshop on human motion analysis and synthesis, held june 26, at Inria in Rennes : http://people.rennes.inria.fr/Julien.Pettre/w26/index.html. This workshop last one full day. Approximately 40 participants attended the workshop.

April 2015: our joint work with UNC on insect simulation is presented at Eurographics 2015 by Weizi Li [Li 2015].

li

March 2015: yearly meeting in Arles, during IEEE VR.

August 2014: common course with UNC on crowd simulation is presented during the SIGGRAPH 2014 Course program.

August 2014: yearly meeting in Vancouver, Canada, during SIGGRAPH 2014.

May-August 2014: Devin Lange, from University of Minnesota, is doing a master internship at Inria with us. His results were presented in a paper written together with G. Ramirez [Ramirez 2014], presented at MIG 2014. 

lange 2014

April 2014: our joint work with UNC on the estimation of crowd simulation parameters to be presented at Eurographics 2014 by David Wolinski.

May 2014: Steve Tonneau is staying at UNC to work on motion planning techniques together with Shonyon Park.

June 2013: We organized a workshop on swarm and crowd motion during the ICPA conference in Estoril, Portugal.

March 2013: Yearly meeting in Orlando, USA (coupled with IEEE VR)

February 2013: Stephen Guy is staying for 2 weeks in our lab.

March-June 2012: David Wolinski stays 4 months for a master internship at UNC.

March 2012: Launch meeting in Los Angeles, USA (coupled with IEEE VR)

Publications

 

[Wolinski 2016] WarpDriver: context-aware probabilistic motion prediction for crowd simulation. D Wolinski, MC Lin, J Pettré. ACM Transactions on Graphics (TOG) 35 (6), 164. 2016

[Park 2016] Dynamically balanced and plausible trajectory planning for human-like characters. C Park, JS Park, S Tonneau, N Mansard, F Multon, J Pettré, D Manocha. Proceedings of the 20th ACM SIGGRAPH Symposium on Interactive 3D Graphics. 2016

[Tonneau 2015] A reachability-based planner for sequences of acyclic contacts in cluttered environments. S Tonneau, N Mansard, C Park, D Manocha, F Multon, J Pettré. International Symposium on Robotics Research (ISSR 2015). 2015

[Li 2015] Biologically‐Inspired Visual Simulation of Insect Swarms. W Li, D Wolinski, J Pettré, M C Lin. Computer Graphics Forum 34 (2), 425-434. 2015

[Wolinski 2014a] Optimization-based pedestrian model calibration for evaluation. D Wolinski, SJ Guy, AH Olivier, MC Lin, D Manocha, J Pettré. Transportation Research Procedia 2, 228-236. 2014

[Wolinski 2014b] Parameter estimation and comparative evaluation of crowd simulations. D Wolinski, S J Guy, AH Olivier, M Lin, D Manocha, J Pettré. Computer Graphics Forum 33 (2), 303-312. 2014

[Ramirez 2014] Optimization-based computation of locomotion trajectories for crowd patches. JGR Ramirez, D Lange, P Charalambous, C Esteves, J Pettré. Proceedings of the Seventh International Conference on Motion in Games, 7-16. 2014

[Bera 2014] Real-time Crowd Tracking using Parameter Optimized Mixture of Motion Models. A Bera, D Wolinski, J Pettré, D Manocha. arXiv preprint arXiv:1409.4481. 2014

[Pettré 2014] New generation crowd simulation algorithms. J Pettré, M Lin. ACM SIGGRAPH 2014 Courses, 4. 2014