While I'm not a fan of the whole "autonomous car" future, and I certainly don't think it will be the future of travel, this is completely wrong. Congestion is often less an issue of theoretical capacity and more an issue of driver error creating ghost traffic jams from obstacle that doesn't exist anymore. A simple example is when a car changes lanes on a congested road, when he does so that usually prompts the driver behind him to stop, and because of disuniform acceleration, that stop creates a chain reaction where every car that passes by that point will slow down which can also reach the all of the cars behind that point, creating a mini traffic jam. If we were to theoretically reach Level 5 automation, the lack of human input would theoretically allow vehicles to run tightly after each other, accelerate uniformly if there is a disruption, and drive in a way that isn't possible by humans without major risk, which will avoid many of the traffic issues we see today, and overall vastly increasing the traffic throughput on roads and highway. The main issue that an automated future has has less to do with what's theoretically possible, and has more to do with feasibility. Any software engineer will tell you that foolproof automation solutions are practically impossible to be 90-95% foolproof, let alone 99-100%, and then you start running into issues such as liability in case of a software failure, and how a driver could take manual control of a vehicle (which if a driver can do that at any point, pretty much kills a lot of the advantages that full automation can bring). Even with stuff like automated trains, there are often cases where control needs to be taken by the central control centre, and this is an environment where all vehicles are on rails with consistent headways which is easy to organize and manage. Doing so with a traffic grid is nigh impossible. As a bonus, Software Engineers are often quite incompetent:
View attachment 307912