Prediction of Pedestrian Speed with Artificial Neural Networks

Final project for the course of Machine Learning in Crowd Modelling and Simulation @ Technical University of Munich (TUM)

Description

This project contains an analysis of the paper Prediction of Pedestrian Speed with Artificial Neural Networks, Tordeux et al., 2018. We tried to replicate their results, as well as to test the approach on new scenarios created with the software Vadere. Finally we discussed weak points of the paper as well as possible solutions/improvements and future developments.
Refer to the report for a more detailed description of the work as well as for understanding the organization of the code.

References

1. Prediction of Pedestrian Speed with Artificial Neural Networks, Tordeux et al., 2018
2. Artificial neural networks predicting pedestrian dynamics in complex buildings, Tordeux et al., 2019
3. Review of Pedestrian Trajectory Prediction Methods: Comparing Deep Learning and Knowledge-based Approaches, Korbmacher and Tordeux, 2021
4. Prediction of pedestrian dynamics in complex architectures with artificial neural networks, Tordeaux et al., 2019