Manon Prédhumeau

PRISAC - Pedestrian Risk Integrated Simulation with an Autonomous Car

PhD thesis of Lydie Baillard - 2025-present

University Grenoble Alps - PACTE - University of Tsukuba

Co-supervision with J. Dugdale and S. Caroly

The project is funded by International Excellence and Outreach (IDEX RI) and involves Grenoble Alps University (PACTE and LIG), IRIT, and the University of Tsukuba, Japan. In dense and dynamic urban environments such as shared spaces autonomous vehicles (Avs) will have to navigate among pedestrians. A key challenge towards the harmonious integration of AVs in public spaces is managing risk perception. This project investigates risk perception and the associated behaviours of pedestrians and passengers during pedestrians-AV interactions. The subject will be addressed from the perspective of the pedestrian at University Grenoble Alps and from the perspective of the AV passenger at University of Tsukuba. The two PhD students will jointly investigate two main research questions: 1. How does risk perception affect the behaviour of the pedestrians around the AV and the passenger in the AV? 2. Can we develop a risk classification scheme linking risk perception to behaviours during pedestrians-AV interactions?

agent-based modelling and simulation, activity-based model, synthetic population

Generation of agent behaviors under environmental constraints. Application to future urban planning scenarios

PhD thesis of Elisa Cueille - 2024-present

University Toulouse Capitole - SMAC research group

Co-supervision with B. Gaudou and N. Verstaevel

Metropolises today are marked by urban growth and densification, leading to significant issues such as congestion, mobility problems, air pollution, and urban heat island effects. Urban planning therefore requires tools such as agent-based modeling and simulation to test the evolution of residents' behaviors in response to different planning choices. However, the design of these models is highly dependent on data that are not always available, and assumes that the environment and individual practices have not evolved since the data collection. Prospective scenarios in long-term future often involve changes in the environment, which require residents to adapt their behavior and activity schedules. This thesis will propose new paradigms so that agents behaviours are no longer purely generated based on past data, and models implementation and exploration to assess the proposed paradigms.

agent-based modelling and simulation, activity-based model, synthetic population

Predicting travel patterns under disruption and change

PhD thesis of Yihang Bai - 2023-present

University of Leeds - School of Geography - LIDA

Co-supervision with E. Manley and Transport for West Midlands

Disruptions, events, incidents, and closures are daily challenges to urban transport systems. Yet, our understanding of how people adapt their travel behaviours in each situation is poorly understood. This project undertakes analyses of changes in travel behaviour following disruption and develop new agent-based models that help to explore the outcomes of future disruptions. The project is in collaboration with Transport for West Midlands, in order to provide real-world application and context for the modelling.

agent-based modelling and simulation, MATSim, GPS data

Modelling and simulation of the mobility needs of the ageing population in the UK and Canada for on demand-responsive autonomous transportation

Research fellowship - 2022-2024

University of Leeds - School of Geography - Leeds Institute for Data Analytics

With E. Manley

By 2030, 11.7% of the world's population is expected to be aged 65 or over and population ageing is particularly strong in Europe and North America. These demographic trends pose challenges to existing mobility infrastructure and services, which play a crucial role in enabling older residents to maintain a high level of independence and well-being. The RAIM (Responsible Automation for Inclusive Mobility) project is a British-Canadian collaboration to address how on-demand autonomous vehicle systems can meet the diverse needs of older populations.

As part of this project, my work contributed to the modelling and simulation of the mobility needs of the ageing population in the UK and Canada, in order to identify the demand for door-to-door transport services.

agent-based simulation, MATSim, Python

Modelling and simulation of realistic pedestrian behaviour in a shared space with an autonomous vehicle

PhD thesis - 2018-2021

Grenoble Alps University - Grenoble Informatics Laboratory, STeamer & Inria, Chroma

With J. Dugdale, A. Spalanzani, L. Mancheva

While autonomous vehicles (AVs) function well on highways and roads i.e. in structured and fairly predictable environments, they are not ready to navigate in shared spaces where segregation between pedestrians and vehicles is minimized. Therefore, AV navigation in dynamic and human populated environments like shared spaces should be addressed. The goal of the HIANIC (Human Inspired Autonomous Navigation In Crowds) project, financed by the French National Research Agency, is to ensure safe and predictable interaction between an AV and pedestrians in shared spaces.

As part of this project, my thesis contributed to the crowd analysis by modelling and simulating pedestrian behaviours in order to give the AV knowledge of its social environment in the present and the near future. The models integrate knowledge from social sciences (proxemics, behaviorism), and use agent-based simulation and robotic functions. Read more...

agent-based simulation, Pedsim_ROS, C++, ROS, Python

SPACiSS simulator: Simulation of Pedestrians and an Autonomous Car in Shared Spaces

Defense slides: here

Smart mobility and adaptive governance

Master 2 internship - 2017

LIRIS - Multi-Agent Systems team.

With V. Deslandres, S. Fenet, S. Pageaud, L. Gervasoni

The SmartGov platform allows the implementation of multi-agent urban mobility simulations and the UrbanSprawl platform uses data from OpenStreetMap to assess urban sprawl in order to limit its negative impact on the cities sustainability. These two platforms shared a common limit: they lacked real mobility behavior data.

My internship had three goals: 1) the creation of a more realistic multi-agent model, 2) the validation of the model with real mobility data on the city of Lyon, and 3) the evaluation of theoretical urban sprawl indicators on the realistic city model.

I added developed a realistic trajectories generator from open data. I used data from traffic sensors on the city of Lyon and Household Travel Survey data to validate the simulated trajectories and evaluate the accessibility indicator. Although perfectible, the mobility behaviors generated are close to the observed ones. Read more...

urban mobility, urban sprawl, multi-agent simulation, Repast Symphony, Java, Python

Design and development of a management tool for wireless sensor networks

Master 1 internship - 2016

Irstea - UR TSCF

With G. De Sousa

The purpose of this internship was the design and the development of a management tool for wireless sensor networks deployed on farms. Each network consists of multiple wireless sensors, spread over a plot, collecting measures such as air temperature or humidity at a given frequency.

I developed a tool to facilitate the management of these wireless sensors. Part of this tool is developed in Java and allows recovering measures taken by the sensors on the field and remotely configuring them. A website with responsive web design, enables centralized management of equipment in stock or on the field, as well as consultation of collected data and remote configuration of sensors. Read more...

web management tool, wireless sensor networks, remote measures, Java, HTML, PHP, CSS, JavaScript Bootstrap, PostgreSQL

Page updated

Google Sites

Report abuse

manon dot predhumeau at gmail dot com

MF205, Manufacture des tabacs