Prototyping Autonomous Vehicle Lane Detection for Snow in VR

Autonomous vehicles (AVs) are gaining momentum, and features such as autopilot are becoming widespread among consumer vehicles. AVs track the road to drive correctly, however, when they fail, the driver must take over. Lane detection is a must-have feature for AVs, which has numerous investigations. Nevertheless, most have gaps in extreme weather, such as snowy winters. The lack of snow datasets adds to this, as most include mild scenarios. This paper presents the prototype of a virtual reality digital twin that will allow training lane detection using synthetic data that would otherwise be difficult to recreate in real life.