Research

The research of the chair for traffic safety and reliability focuses on system and model-based reliability analysis, multi-agent systems, interacting particle systems, pedestrian and evacuation dynamics, highway and mixed urban traffic, intelligent transportation systems, and advanced driver-assistance systems.

We develop and analyse mathematical models and numerical simulation tools based on differential systems, stochastic processes, or data-driven algorithms. We calibrate and train models and algorithms using experimental and field data. The aim is to understand, predict, or control complex systems, including collective dynamics. Computations are performed using open source packages in R or Python and multi-agent simulation platforms such as NetLogo or SUMO.

Approach: Modelling → Analysis → Simulation → Optimisation

Duality: Linear models ↔ Non-linear models
Physics and knowledge-based ↔ Data-driven / Deep learning
Microscopic approaches ↔ Mesoscopic / Macroscopic approaches

The chair strongly supports the principles for open science, emphasising open source development, reproducibility, and the FAIR (Findability, Accessibility, Interoperability, and Reusability) principles.

Focus 1

Traffic flow modeling & driving automation

Focus 2

Pedestrian & evacuation dynamics

Focus 3

Reliability engineering & measurement methods

Pedestrian trajectories

Dataset collection at Forschungszentrum Jülich – Experiments
OpenTraj – Naturalistic
UniD dataset – Vulnerable users, naturalistic
Trajnet++ trajectory forecasting challenge – Naturalistic
PeTrack – Open-access pedestrian tracking software

Road vehicle naturalistic trajectories

NGSIM dataset – American highways und urban roads
Ubiquitous Traffic Eyes – Chinese highways
pNEUMA dataset – Congested urban traffic
HighD dataset – German highways
InD dataset – German urban intersections
RounD dataset – German roundabouts
OpenACC Database – Trials with vehicles equipped with ACC systems

Predictive maintenance

Kaggle data set – Real & synthetic data
NASA Prognostics Center of Excellence Data Set Repository – Real & synthetic data
UCI Machine Learning Repository – Real & synthetic data
Bearing data set – Real data
Air compressor datasets – Real data

NetLogo – Simulation of multi-agent systems (logo)
JuPedSim – Simulation of pedestrian and crowd dynamics (python)
Vadere – Simulation of pedestrian and crowd dynamics (java)
Umans – Simulation of pedestrian and crowd dynamics (C++)
GAMA Platform – Agent-based modelling and simulation (java)
Eclipse SUMO – Simulation of Urban MObility (C++)
CARLA Simulator – Autonomous driving simulator (python, C++)
MATSim – Large-scale agent-based transport simulations (Java)

Traffic flow and pedestrian dynamics

Complex systems and multi-agent systems

Intelligent transportation systems and automated vehicles

Reliability engineering

Artificial intelligence and machine learning

wikicfp.com (conferences and calls for paper search engine )

Collective Dynamics is a diamond open-access multidisciplinary journal publishing research in the field of pedestrian dynamics, vehicular traffic, and other systems of self-driven particles or interacting agents.

The journal covers, but is not limited to, the following areas:

Highway, urban, and bicycle traffic
Pedestrian and evacuation dynamics
Granular flow and dynamics of granular materials
Biological systems: social insects, animal groups, cell and bacterial migrations, swarm, flocking and schooling dynamics, …
Non-living systems and active matter: swarm robotics, molecular motors, micro-swimmers, rods, interacting particle systems, …
Complex systems, complex networks and their dynamics: multi-modal transportation networks, multi-agent systems, internet, epidemics, social networks, opinion formation, …