Towards an optimal right-turn assistant system to avoid accidents with vulnerable traffic participants

October 16, 2007, the German Road Safety Council (Deutscher Verkehrssicherheitsrat e. V.) decided to start basing their recommendation on the Vision Zero methodology. Although a lot has already been done to make the roads safer for unprotected traffic participants there is still a potential for improvement especially in avoiding accidents with cyclists. The European Commission has already decided that the obli-gation of having the blind-spot assistant systems will gradually include more and more vehicle types as well as retrofitting older trucks and buses. In this paper, we are evaluating how different sensors may contribute to an optimal turning assistant system besides the well-known existing solutions which typically rely on one kind of sensor. Moreover, we focused on a solution, which includes sensors, that have not yet been used for this purpose. In that sense, we are using the advantages of sensor-fusion techniques together with object classification by using machine learning algorithms. Our research spans from the more conventional choices like cameras and 77GHz radar to less apparent such as Real-Time Location System or bicycle bell sound recognition neural networks. Besides that, we decided to not only warn the truck driver but also the unprotected traffic participants about a possibly dangerous situation, which is as well a novelty to our best knowledge. Since our work is still in progress, especially due to the huge amount of training data, there are no final numbers regarding accuracy that could be given. Nevertheless, we would like to share what we have done thus far.