I disagree. A visible-light based system like Tesla proposes is, quite frankly, a joke. Camera-based systems are too easily confused and require too much processing power to work in high-speed situations. (For example, at least one of the Tesla deaths was the result of the camera system failing to identify that a stopped object wasn't part of the road. A radar-based system would have easily recognized the obstacle.) Camera systems are also too dependent on maintaining the cleanliness of the sensors, and their performance degrades based on lighting conditions. While they certainly will be part of future AV systems (i.e., for street signs), they won't be the primary sensor.

Some combination of radar will form the basis of AV systems going forward. LiDAR is the most likely foundation, given its accuracy and speed, especially as LiDAR is expected to drop in price by nearly 99% over the next 5 years. (Recent developments in the underlying technology, plus economies of scale, have reduced the price of a full LIDAR suite from nearly $100,000/vehicle to just over $1000/vehicle. Technologies in the pipeline that are close to commercial production should reduce that cost to $100/vehicle or less.)