Road and free space detection (with or without markings) by Computer Vision and Deep Learning

 
 

. Detects road shapes with or without markings and also detects surface of free space. ADAS and Autonomous Vehicle need to know where the road is and if it is drivable     . Computes size of free space and sends it to SafetyNex. Small free space is a factor of driving risk

 

---

 

software component

Image(t)

Road(t) : Coordinates of road borders Coordinates of free space borders (even if there are no markings) In the image space in pixels In the real 3D world in meters

 

---

 

NEXYAD algorithm RoadNex

Pan in real time on a smartphone (pan is the most complex movement for detection systems)

 

Cross a round about (very complex use case for autonomous shuttles markings are out of field of vision)

 

Follow a desert track (unstructured road and NO markings)

 

Follow a road (simply a must have but runs on a regular smartphone instead of heavy computing system)