Current position: Postdoctoral Associate, ITSLab, Massachusetts Institute of Technology under the Singapore MIT Alliance for Research and Technology (SMART)
Thesis (2014): Probabilistic Safety Analysis using Traffic Microscopic Simulation
PhD Supervisors: João Cardoso, LNEC; Moshe Ben-Akiva, MIT; Filipe Moura, IST.
Research topic/area: Intelligent Transport Systems, Road Safety, Risk Analysis (SCUSSE project)
Links: Personal webpage
Thesis Abstract: Traffic microscopic simulation applications are currently a common tool in road system analysis and several application attempts to safety performance assessment have been recently carried out. However, current most common approaches still ignore causal relationships between different levels of vehicle interactions or accident types, lacking for a physical representation of the accident phenomena itself. A new generic probabilistic safety assessment framework for traffic microscopic simulation tools is proposed. The probability of a specific accident occurrence is assumed to be estimable by an accident propensity function, composed by a deterministic safety score component and a random component. The formulation of the safety score component may be specified depending on the type of occurrence and on the simulation features. The generic model is then specified for the case of urban motorways for no-accident events and three types of accidents: rear-end, lane-changing and run-off-road accidents. To deal with the lack of available trajectory data for different occurrence types, artificial trajectories from a calibrated microscopic simulation tool are used. These trajectories are obtained following a comprehensive calibration effort: extracting trajectories for a generic scenario, calibration of the simulation tool using the collected trajectories, and re-calibration of the simulation model using aggregate data for each event selected at replication. An advanced method for automatic extraction of vehicle trajectories using aerial imagery is presented, in order to collect the detailed traffic variables. A global sensitivity analysis based calibration is proposed to deal with uncertainty in the detailed calibration of complex models. The parameters of the safety model are estimated using artificial vehicle trajectory data calibrated for the Portuguese A44 motorway and using the MITSIMLab simulator. With this study it is shown how traffic microsimulation tools may replicate detailed traffic statistics that are essential to explain different accident phenomenon and how the quality of this replication is strongly linked to the simulation modelling formulation, the calibration methodology and the available data.
Keywords: traffic microscopic simulation; road safety; probabilistic assessment; driving behaviour modeling; surrogate safety measures; discrete
 “Dealing with uncertainty in detailed calibration of traffic simulation models for safety assessment”. Lima Azevedo, C., Ciuffo, B., Cardoso, J., Ben-Akiva, M. E. Transportation Research Part C. Forthcoming.
 “A sensitivity-analysis-based approach for the calibration of traffic simulation models”
Ciuffo, B., Lima Azevedo, C. (2014), IEEE Intelligent Transportation Systems Transactions, Volume 15, Issue 3, pp. 1298 – 1309.
 “Vehicle tracking using the k-shortest paths algorithm and dual graphs”
Lima Azevedo, C., Cardoso, J., Ben-Akiva, M. E. (2014) Transportation Research Procedia, Volume 1, Issue 1, 2014, pp 3-11.
 “Road safety performance indicators for the interurban road network.”
Yannis, G., Weijermars, W., Gitelman, V., Vis, M., Chaziris, A., Papadimitriou, E., Lima Azevedo, C. (2013), Accident Analysis & Prevention, Volume 60, November 2013, Pages 384–395.
 “Automatic vehicle trajectory extraction by aerial remote sensing”
Lima Azevedo, C., Cardoso, J., Ben-Akiva, M. E., Costeira, J.P., Marques, M. (2013), Procedia – Social and Behavioral Sciences, Volume 111, pp 849-858.