The danger of motorcycle accidents is ubiquitous during the otherwise relaxing and enjoyable activity of riding a motorcycle. The consequences can be severe and the economic burden, both on the individual and the state, is high. Yet, when wanting to prevent such accidents, it can be seen that they are hard to predict, due to the high complexity of individual factors playing a role in each single accident.

To tackle this issue and extract generalizable characteristica of driving dynamics, the authors present the findings of “viaMotorrad”, a project to obtain motorcycle dynamics data on selected roads in Austria and determine the risk of an accident at given road sections. This is a collaborative project by the Austrian Road Safety fund, between the partners Austrian Institute of Technology, TU Wien and KTM.

Through the use of supervised machine learning techniques we demonstrate that there are indeed generalizable factors in the driving dynamics at previous accident sites and use these factors to determine further critical road sections. These results are a first step towards an objectification of motorcycle driving risk and semiautomated risk assessment of roads for motorcycle riders. The method offers the possibility of identifying critical road sections through analysis of a small number of test drives.