Road traffic is highly complex and difficult to describe. And yet it is absolutely essential that we manage to present it in a compact form in meaningful test scenarios. Only in this way can we succeed in bringing highly automated driving functions to series production readiness in the foreseeable future.
The test scenarios themselves can be generated either logically, i.e. from abstract considerations, or data-driven, i.e. from analyses of the traffic situation. Or a combination of these two possible approaches. And it is precisely for this purpose that it is important that we use as much information and knowledge as possible to generate test scenarios that are as relevant as possible: We have to merge different databases, such as data from traffic observations and accident records. You can read how this can be done here: “With accident and traffic observation data to the optimal test scenario”.
Almost every new car today is equipped with a number of different driver assistance systems, whether relatively simple lane change warning systems or automatic emergency brake assist systems, which are designed to prevent accidents in urban traffic or mitigate collisions. In addition, autonomous driving, especially on motorways, is already within sight and will come in the (near) future.
The question that inevitably arises for many engineers is the influence of these systems on real road traffic. To what extent is the increasing market penetration noticeable in traffic and to what extent do the systems prevent accidents and injuries? The latter question aims at the pre-crash phase, i.e. the time before the first collision of an impending accident – a question that cannot be answered immediately due to a lack of data in real road traffic.
To address these questions, an international consortium, P.E.A.R.S (“Prospective Effectiveness Assessment for Road Safety”), was founded in 2012. It consists primarily of partners from the automotive industry, scientific institutions and universities (e.g. TU Dresden). The aim is to develop an understandable, reliable and accepted simulation method for the quantitative evaluation of accident-preventing technologies / driver assistance systems. In addition, an ISO standard for the predictive evaluation of the road safety of active driver assistance systems installed in the vehicle is to be developed – also here with the inclusion of virtual simulation methods, which should enable a cost-effective evaluation.
The P.E.A.R.S. consortium is divided into a total of four sub-working groups, whereby since November of last year I have been allowed to participate in Working Group A (WGA) “Methods, models and effectiveness calculation”. The main aim of WGA is to answer the following three questions:
- Definition of evaluation targets and a basis for comparison (“baseline”) for evaluation
- Implementation of virtual simulation models
- Development of metrics for estimating the effectiveness of safety systems
Since I am still in the process of getting an overview of the already extensive work of working group A, I cannot define my exact topic at this point yet. However, it seems that in the WGA – similar to SePIA – I will be dealing with the comparability / extrapolation of simulation results for other regions / countries.
So far the work in this extremely international consortium gives me a lot of pleasure and I am looking forward to report already in March from the General Assembly in Frankfurt. For further information / future publications, a regular visit to the P.E.A.R.S homepage at www.pearsinitiative.com is definitely recommended!