top of page
  • Alexandre

Stage de recherche (physique + IA): traversée d'une foule par un véhicule autonome

Proposition de stage de recherche à l'Institut Lumière Matière, institut public affilié au CNRS et à l'Univ. Lyon 1, pour un niveau Master:

Context. Autonomous cars have started to appear on the streets of several cities across the globe and may become the dominant form of urban transport by 2040 [1]. So far, most research has focused on their interactions with the built environment and other cars in their vicinity, but interactions with pedestrians are also a subject of major concern for public safety, for sparse pedestrians [2] as well as for denser crowds. Navigation through dense crowds has recently started to attract academic interest [3-4], but these endeavours are faced with the intricacy of modelling the response of such crowds. In particular, reproducing the crossing of a static crowd by an intruder is still a challenge for models of pedestrian dynamics [5-7].




Goal. The research internship is aimed at unlocking this situation by modelling admissible and non-admissible trajectories of a (simplified) autonomous car through a crowd. More precisely, agent-based models will be exploited and further developed to simulate a realistic crowd’s response to the traversing motion of an autonomous vehicle. More precisely, the intern will tackle the following problem : Assuming that the car follows a given trajectory or obeys simple equations of motion, is there a risk of collision with pedestrians in the crowd ? Can one delineate the crowd's responses that are admissible (i.e., those that do not lead to any collision) and those that are not?

To this end, the intern will

  • make use and further develop agent-based models for pedestrian motion

  • develop a theoretical method to delineate admissible crowd’s responses, notably by putting forward quantitative indicators to gauge how acceptable a crossing is

  • contribute to the development of a 3D visualisation tool to illustrate the output of the model.


Profile and skills. We are looking for a motivated and autonomous intern

  • with a solid background in Physics (Complex Systems and/or Statistical Physics and/or Condensed Matter Physics)

  • with a good grasp of numerical tools and programming (ideally, C++ and Python)

  • Previous experience with 3D modelling tools (like the Unity Platform) would certainly be an asset, but is not a requirement in any way.


The intern will be co-supervised by Alexandre NICOLAS (Institut Lumière Matière) and Olivier SIMONIN (Citi-lab) and will be based in one of these two labs. The project takes place in the frame of a joint programme funded by Fédération d’Informatique de Lyon (CROSS).



References [1] Cugurullo, F., Acheampong, R. A., Gueriau, M., & Dusparic, I. (2020). The transition to autonomous cars, the redesign of cities and the future of urban sustainability. Urban Geography, 1-27 [2] Poibrenski, A., Klusch, M., Vozniak, I., & Müller, C. (2021). Multimodal multi-pedestrian path prediction for autonomous cars. ACM SIGAPP Applied Computing Review, 20(4), 5-17. [3] Bresson, R., Saraydaryan, J., Dugdale, J., & Spalanzani, A. (2019, June). Socially Compliant Navigation in dense crowds. In 2019 IEEE Intelligent Vehicles Symposium (IV) (pp. 64-69). IEEE;

[4] Prédhumeau, M., Mancheva, L., Dugdale, J., & Spalanzani, A. (2022). Agent-Based Modeling for Predicting Pedestrian Trajectories Around an Autonomous Vehicle. Journal of Artificial Intelligence Research, 73, 1385-1433. [5] Nicolas, A., Kuperman, M., Ibañez, S., Bouzat, S., & Appert-Rolland, C. (2019). Mechanical response of dense pedestrian crowds to the crossing of intruders. Scientific reports, 9(1), 1-10.

[6] Bonnemain, T., Butano, M., Bonnet, T., Echeverría-Huarte, I., Seguin, A., Nicolas, A., ... & Ullmo, D. (2023). Pedestrians in static crowds are not grains, but game players. Physical Review E, 107(2), 024612. [7] Echeverría-Huarte, I., Roge, A., Simonin, O., & Nicolas, A. (2023). Foundations of continuous agent-based modelling frameworks for pedestrian dynamics and their implications. arXiv preprint arXiv:2309.12798.


Comments


logo_CNRS.png
logo_ILM.png
logo_univ_Lyon.png
bottom of page