Jury Composition :
rof. Said MAMMAR | Rapporteur | Université d'Évry Paris-Saclay, France |
Prof. Giandomenico CARUSO | Rapporteur | Polytechnic of Milan, Italy |
Prof. Olivier SENAME | Examinator | Université de Grenoble Alpes, France |
Prof. Patricia DELHOMME | Examinator | Université Gustave Eiffel, France |
Dr. Michel BASSET | Examinator | Université de Haute Alsace, France |
Prof. Andrea Simone | Invited member | University of Bologna, Italy |
Dr. Hocine Imine | Thesis Director | Université Gustave Eiffel, France |
Prof. Claudio LANTIERI | Thesis Codirector | University of Bologna, Italy |
Prof. Francesca DE CRESCENZIO | Thesis Cosupervisor | University of Bologna, Italy |
Abstract:
While cycling as a mode of transport is becoming increasingly popular, cyclists remain among the most vulnerable road users due to their exposure to safety risks and accidents. Assessing these risks and understanding cyclist behavior are essential for improving cyclists' safety and infrastructure design.
This thesis focuses on risk assessment and cyclist behavior by developing a Risk Index (RI) to quantify safety risks across different road segments. This index integrates main factors, including infrastructure characteristics and geometries, cyclist behavior, and external conditions such as traffic and weather. This part followed a multi-phase methodology, including risk factor identification, risk weighting, and risk index formulation.
The second part of the research was based on the evaluation of the risk index using two experiments: a bicycle simulator to assess risk level on simulated road segments and an instrumented bicycle to conduct a real experiment in Bologna, Italy, and assess risk level on different urban road segments. The thesis analyzed the correlation between the RI and subjective feelings of participants to evaluate its reliability, and the methodology used. Furthermore, cyclist behavior, such as gaze metrics and bicycle kinematics, was studied to understand how cyclists interact with different road users and road environments such as intersections, shared roads, crossings, and dedicated cycle paths.
Since sharing cycling infrastructure with pedestrians increases potential conflicts, virtual reality (VR) was used to analyze pedestrian risk perceptions in segments involving interactions with cyclists. A bicycle path from the bicycle simulator was reconstructed in VR and evaluated from the pedestrians' perspective. Comparing cyclist and pedestrian perceptions of shared paths leads to obtaining mutual safety concerns.
Finally, the thesis developed theoretical active safety systems for cyclists, including steering and speed control systems using PID and sliding mode control to reduce lane departures and excessive speeds. Cyclists’ acceptance of these safety assistance tools was also evaluated.
This thesis follows risk assessment methodology for bicycle safety, understands cyclist behavior, and contributes to improving cycling infrastructure planning, promoting safer interactions, and providing active safety solutions to enhance cyclist stability and reduce accidents.