Let's talk about the future of autonomous driving

I increasingly feel like I've reached an age where I'm caught in a dilemma. My dreams seem to be getting further and further away, so I have to resort to indirect methods. The number of unresolved issues is growing, and I'm facing more and more problems, forcing me to reluctantly give up some things. Student days were so much better!

I've recently started reading again, something I haven't done much since going abroad. I do carry my reading notes with me, but the last time I wrote anything was in 2013. I recently read an article about autonomous driving ("The Day Autonomous Driving Finally Arrives Will Completely Change Our Travel Methods"). The article mentioned some data and prospects. I've discussed this with friends before, and there are many points that weren't mentioned in the article, so I'm taking this opportunity to write an article to share my thoughts.

Let's set aside the question of when autonomous driving will arrive, but it is inevitable. Why? Looking at global development, simple, repetitive tasks are gradually being replaced by machines—examples abound. Furthermore, from a human perspective, people avoid doing things themselves whenever possible, and prefer simple solutions to complex ones. However, driving is considered by many to be an extremely complex "labor"—requiring not only keen observation and quick thinking but also quick reflexes. And once a problem occurs, it can cause irreparable damage. Imagine a car losing control and careening recklessly down the road, or misjudging a situation and injuring someone—nobody wants that to happen. As a layperson, I largely agree with this viewpoint, but even so, autonomous driving may not be impossible.

What would roads look like if autonomous driving were achieved? We need to distinguish between two scenarios: roads with fully autonomous vehicles and roads with a mix of human and autonomous driving. The former would be far safer than the latter because, through communication between vehicles, a complete road network topology could be formed, giving each vehicle a comprehensive view of the road network. In this scenario, if an emergency occurs, all vehicles could make the most correct and unified action. In the latter scenario, due to the unpredictability of human driving, autonomous vehicles might not be able to fully anticipate or predict human behavior in special circumstances, potentially leading to accidents.

Based on this idea, once autonomous vehicles are developed, a dedicated lane will inevitably be found in the transportation system. Over time, as human-driven vehicles are phased out, the dedicated lane will become the main road, and eventually, human-driven vehicles will become a dedicated lane. Some people think that this will be very costly to road reconstruction, but it's not. Just like the current dedicated bus lanes, it only requires designating a specific area. How long will this take? Assuming that autonomous vehicles can be mass-produced by 2030 (as predicted in the text), without special policy guidance, it will take about 10-20 years for purely human-driven vehicles to cease production, just like when automatic transmissions replaced manual transmissions. Vehicles will default to autonomous driving, and human driving will need to be switched off. That's 2050. By 2050, we will still need another 10-20 years to phase out those new cars (cars purchased in 2049). Around 2060-2070, autonomous vehicles will have an absolute numerical advantage in the market. At this point, it can be understood that the main road will serve autonomous vehicles.

Regarding safety, the apparent loss of human control over vehicles can be unsettling, but a closed-loop autonomous driving system would alleviate this concern. This is analogous to the small, fully electric transfer tracks at airports, where vehicles move independently (quite absurd!). In this scenario, all vehicles can be viewed as a unified system traveling on the road. Traffic congestion would also be resolved. Many current "phantom" traffic jams arise from lane changes, exits, and merges, caused by vehicles being unaware of each other's movements. With an autonomous driving network, this situation could be entirely managed through machine calculations, as all vehicles would know the next move of every other vehicle.

Under this assumption, what would happen if autonomous driving were achieved by 2070? What would be the next step? We know that with the development of information technology, humans are acquiring information at an increasingly faster pace, leading to a faster rhythm of life and making time increasingly precious. In Chinese cities, public transportation is still available, but in the US, almost everyone drives to work. On average, people spend an hour and a half driving every day, a time that is completely unnecessary but unavoidable. Imagine the future of Uber and car-sharing. If autonomous driving is achieved, why would we still need to own cars?

Currently, I drive to work, park in the garage, and then drive back and park in the garage again after get off work. Aside from commuting, the car is essentially idle. Another example: I need to go to a gathering in the city, but parking is difficult. In this case, I'd consider public transport or a taxi, missing out on the convenience of owning a private car. This problem will be completely solved once autonomous driving is realized. Imagine all roads becoming like railway tracks. Where you want to go, you simply wait for the high-speed train, hop on, and hop off at your stop. Long waiting times or no cars available? Probably not. Imagine everyone owns a car now, but the usage rate is probably less than three times a day. If all these cars were put into this operation, there would be a surplus, not a shortage. The cost would also be far lower than private cars, making it more like taking public transport.

I look forward to such a future, but I guess I won't be able to walk anymore by then.

P.S.: I'm not from this field. I just talked to someone about it before, looked at some things, and rambled on a bit. Please point out any shortcomings.