F2008-12-018
Evaluation of Safety Effects of Speed Assistance Systems by Simulation: A Critical Review
Speed is a crucial issue of road traffic safety. In-vehicle speed assistance systems could affect general mean speed and speed variance, therefore, increase road traffic safety. The purpose of this paper is to contribute to the knowledge concerning the evaluation of safety effects of speed assistance systems by simulation modelling. The paper reviews the state-of-the-art traffic flow models based on literature study. In addition, it discusses limitations of current driving behaviour models. Also, simulation models specifically applied for studying safety effects of speed assistance systems are analysed. Furthermore, the paper addresses some issues concerning safety simulation for discussion, and proposes topics for further research on speed assistance systems.
There is a need to study the traffic flow process and the interaction between humans, vehicles and road infrastructure (subject to legislation and traffic regulations) by simulation model. The review of the current traffic flow models and their application to driving assistance systems shows that microscopic and macroscopic models are not equally suited for all types of traffic flow analyses with vehicles equipped with driving assistance systems. Mesoscopic (or kinetic) models have more parameters with a direct relation to driver behaviour that can be influenced by the traffic control policy (e.g. desired speed distribution), compared with pure macroscopic models. Safety simulation is a rather difficult, if it is not impossible. The complexity is especially due to limited knowledge on human behaviour and available (observable) data. None of current simulation models can fulfil the task of evaluation of safety effects. Previous research on evaluating safety effects of speed assistance systems are mainly based on microscopic traffic simulation. Most assumptions in these researches are rather arbitrary, unrealistic and coarse. Therefore, the value and credibility of the results of these studies results are doubtful. In addition, clear weakness exists, to some extent, in the research design of these studies, which opens room for arguments. More fundamental behaviour studies are prerequisites for safety simulation. Until now, the use of mesoscopic model principles for the evaluation of speed assistance systems has not been explored. The challenge is to implement the effects of traffic flow control and driving assistance systems into these models by flexibly adapting the microscopic model foundations, after which the behavioural changes can be transcribed into a macroscopic model formulation.
Poster presentation: Simulation and testing