Putting aside any media outlet’s take on this, yes he’s wrong but not for the reasons most fans and “futurists” would likely consider. First and foremost, any safety critical system must have backups built in. Backup sensing technology will hopefully be a required component in any future authority approved system. Sadly, I expect Europe will once again lead in this area and the US/Canada will lag far behind.

But just the technology of lidar itself is vastly superior to visible light camera systems in many instances. Different waves can peer through fog and smoke, high quality DSPs can filter out things like precipitation, lidar brings the light it needs to operate while cameras rely on ambient light or headlights. Lidar also receives speed and distance data as results from point clouds without any need for motion estimation.

It’s also likely that radar will be a requirement in any serious system, even thought lots of people point out some older arguments against radar. The fact of the matter is, even thought it’s low resolution because of wavelength, repeated measurements can increase the effective resolution. And again, radar brings the energy it requires for its operation rather than relying on ambient conditions, it sees through precipitation, fog, smoke, it can be used at greater distances than visible light or even lidar systems.

I also think high quality IR cameras would be an amazing addition to any sensor suite because they can detect objects approaching the edge of the road like animals coming out of fields and forests. Something that kills people somewhat frequently, and visible light cameras have a hard time with at night.

Basically, the number of additional sensors isn’t actually a detriment to automated or assistive driving packages, they are a benefit. Attempting to produce the least expensive system by removing sensors here is equivalent to building a submersible vehicle out of carbon fiber. We know the result before anybody ever attempted it.