Master Thesis on “Anomaly detection in people flow through maps of dynamics”
Supervisor: Prof. Ville Kyrki (ville.kyrki@aalto.fi)
Advisor: Dr. Francesco Verdoja (francesco.verdoja@aalto.fi), Prof. Tomasz Kucner (tomasz.kucner@aalto.fi)
Keywords: mapping, maps of dynamics, anomaly detection, people flow, statistical methods
This thesis is part of an on-going collaboration with KONE Oyj
Project Description
People movement within a space can be modeled through the use of Maps of Dynamics (MoDs), i.e., probabilistic spatial representations modeling the local flow of people within an environment. MoDs have wide applications ranging from social-aware robotic navigation to human motion prediction.
Goal of this thesis is to explore whether MoDs can be used to detect anomalous behaviours in people flow. The ability to detect significant shifts in how people move in public spaces could be used to recognize situations of risk and to quickly respond to emergencies.
KONE Oyj, partner in this thesis, has collected months of human trajectories within Helsinki Central Station (Rautatientori) and the data collected, made available to us in a AWS repository, will be used to validate the proposed method.
Deliverables
- Review of relevant state-of-the-art literature;
- Implementation of a relevant MoD from literature and its use to model the data from KONE;
- Design and development of an approach for anomaly detection in people flow leveraging the implemented MoD;
- Evaluation of system performance on real-world data.
Pre-requisites: python/C++ (high), statistical methods (medium), Linux (medium), AWS (low)
Start: Available immediately