The accident statistics of the European Road Safety Council (ETSC) shows that 47% of people killed in road traffic accidents in 2018 were pedestrians, cyclists or motorcyclists. 83% of cyclist fatalities were caused by a collision with a motorised vehicle.
In the Bike2CAV project, AIT is researching with its partners BikeCitizens, Boréal Bikes, Kapsch TrafficCom, KFV, Salzburg Research and the University of Salzburg new methods to increase the road safety of cyclists. The focus of the AIT research group Assistive & Autonomous Systems is on methods to reliably perceive cyclists in different traffic situations. The particular challenges here are that "cyclists show a high degree of variation in their appearance. Also different weather conditions make reliable recognition difficult," explains Martin Fletzer, head of the project group at AIT.
The scientists are drawing upon their research in the field of autonomous vehicles and cooperative robots. Here, too, integrated systems should recognise and understand the cyclist's intention as early as possible. Hand signals play a decisive role, as they indicate a possible change of direction by the cyclist. "Acutally, we are researching the recognition of hand signals with neural networks. With our new approaches from the field of Deep Learning, we have already been able to achieve initial successes and could determine the direction in which cyclists are going to move. At the end of the project in 2023, we will evaluate the results on the basis of selected, public road intersections in order to qualify the increase in road safety for all parties involved," says Martin Fletzer.
die Presse 18/02/2021, p.14 is covering the story.