It seems to require a lidar reflective object. Likely more generally, the effectiveness lowers the less objects there are to bounce and return signal.
It could probably work with less accuracy/resolution against visible vehicles in the opposite lane, a hedgerow, postal box, pedestrian carrying a visible laptop and possibly synthesize all of these to improve its guess.
It seems to require a lidar reflective object. Likely more generally, the effectiveness lowers the less objects there are to bounce and return signal.
It could probably work with less accuracy/resolution against visible vehicles in the opposite lane, a hedgerow, postal box, pedestrian carrying a visible laptop and possibly synthesize all of these to improve its guess.
The video thumbnail implies bouncing off the ground, not a wall. Not sure how the geometry works out for that