Institute for Transport Studies (ITS)

Driver-oriented virtual testing of vehicle safety and automation

Supervisor: Dr Gustav Markkula (possible co-supervisors Professor Natasha Merat and Professor Richard Romano)

As vehicle systems for automation and driver support become increasingly complex, so do also their verification and evaluation. To ensure that vehicle automation features are robust and error-free requires test driving over such large distances that it becomes impossible to carry it all out in physical reality for every hardware or software update. Likewise, to estimate the actual impact of in-development systems on very rare externally arising safety-critical situations is also more or less infeasible without taking recourse to virtual methods. This project will aim to improve methods for virtual testing of safety support and automation functions, with an emphasis on (1) completely computational, model-in-the-loop testing, with (2) faithful representations of human driver behaviour, when applicable.

The first step of the project will be to agree the exact scope in terms of addressed systems and traffic scenarios. This will be determined with up-to-date input from vehicle industry, to which the research group has strong ties. Based on the agreed scope, industry co-supervision might be possible. The researched methods for virtual testing could include so called "what-if" simulations based on naturalistic crash data, and/or large-scale Monte Carlo simulations. Furthermore, situations of varying criticality should ideally be addressed, as well as situations arising both externally from surrounding traffic or internally from e.g. functional failure.

A specific current strength of the research group in this domain is the understanding and modelling of human behaviour in safety critical situations, and this should be leveraged and ideally developed further, if appropriate supported by additional data collection with human participants.

The suitable candidate will have a background in Engineering or similar quantitative discipline, with an interest in and prior working knowledge of at least some of the following: programming (esp. MATLAB), data analysis, signal processing, control theory, vehicle dynamics, complex systems, human behaviour and psychology, neuroscience.

Suggested reading:

Markkula, G. (2015). Driver behavior models for evaluating automotive active safety: From neural dynamics to vehicle dynamics. PhD thesis, Chalmers University of Technology.

Markkula, G. Engström, J., Lodin, J., Bärgman, J., Victor, T. (2016). A farewell to brake reaction time? Kinematic-dependent brake response in naturalistic rear-end emergencies. Accident Analysis & Prevention, 95A, 209-226.

Zhao, D., Huang, X., Peng, H., Lam, H., LeBlanc, D. J. (2016). Accelerated Evaluation of Automated Vehicles in Car-Following Maneuvers. Submitted for publication. arXiv:1607.02687

Search site