Lidar versus visual processing for automated driving

Lidar, or laser-based radar, is a key enabling technology for most auto makers’ self-driving cars. Lidar is good at identifying objects far down the road and determining speed. Unlike traditional cameras, the technology doesn’t get blinded coming out of tunnels or in other tough driving conditions. For example, the autonomous taxis that operate in Phoenix and Las Vegas are equipped with lidars, in addition to cameras and traditional radars.

Source: Lidar Needs to Get Cheaper and Better. Why This Stock Could Benefit. | Barron’s

Tesla has relied on visual processing systems rather than lidar. Lidar has, historically, been expensive.

I suspect the Lidar versus vision processing is going to be similar to what occurred in photography for automatic focus systems. In photography, much of the camera world went with phase detect auto focus while Olympus and Panasonic stayed with contrast detect auto focus, or Panasonic’s newer tech, Depth From Defocus (DfD). Indications are that eventually, most cameras will use phase detect technology (in addition to contrast detection).

I wonder if future self driving vehicles will end up using both lidar and image processing. I think that is where it is headed.