Road detection RoadNex may help to avoid TESLA accidents

Most camera-based road detection systems are based on detection of white lines. Thoses detectors are very efficient if white lines exist (it is not the case on every road in Europe) and if they are "clean" and not tricky.

For instance, if you have a construction zone, you keep the white line detection on the right side, you lose it on the left side (so you may think
that they are erased on the left side because it happens all the time). Plots (or temporary separation wall) used to delimiter the construction zone are too "small" to be kept as obstacles (because camera-based obstacle detection is looking for pedestrians and cars) ... the radar is filtered not to see immobile objects (dolpler filter) because immobile objects (the road itself evey time the slope change, bridges, side objetcs every time you have a curve, ...) would bring too manu echos.

tesla accident scheme

 

This kind of system leads to an accident if the driver doesn't disable the sel-driving mode.

On the contrary, the road detection module RoadNex detects borders of drivable area and it could have avoided a recent accident with a TESLA. It is not a matter of performance : TESLA detection systems are very efficient. It s a matter of functionalities ...

The real question is : "what is a road" ?

Here is a picture of what RoadNex detects on this recent accident :

Tesla accident 02

 

RoadNex detects the borders of drivable areas ... and those borders are NOT white line although it is a highway scene.

RoadNex may also verify (with a second algorithm) that the surface is drivable.

See demos : http://nexyad.net/Automotive-Transportation/?page_id=412