With autonomous driving, cargo could be transported by very small vehicles. Very small vehicles could use very small tunnels. Given the expense of tunnelling, very small tunnels could be much cheaper to build than larger tunnels. This could be useful in urban areas, to avoid road traffic, allow cargo to be transported late at night without creating too much noise, and use an electrified system (rail, cable, or trolley) rather than use relatively heavy and bulky batteries or internal combustion engines. It could also allow autonomous vehicles to avoid (or reduce) sharing the road with human-driven vehicles, and could allow for autonomous driving without requiring expensive systems like Lidar and without having to deal with technical challenges like snow.
Such tunnels would also facilitate the use of shortcuts, whether within cities (e.g. to pass under a highway or river valley) or to cross natural barriers in order to reach cities. This would be especially useful if the price of oil (and/or the price of energy in general) were to increase. Cargo often takes a lot of energy to transport, but by using shortcuts you reduce the total travel distance required. Plus, by transporting more goods at night, you can benefit from power prices often being much cheaper at night.
If these tunnels are going to be built, the next obvious question becomes “could humans travel in them too?” This question has already become popular, of course, with Elon Musk’s “hyperloop” concept and Boring Company being the most obvious example.
Obviously, though, there are challenges to transporting humans when compared to cargo. First, there are safety concerns. Second, there is comfort: plans like Hyperloop assume that people don’t want to travel around lying on their backs, which means that tunnels to transport people would have to be bigger (and thus more expensive) than those used to transport cargo. Third, there is speed. Plans like Hyperloop assume that humans want to get from one place to another quickly. But with increased speed comes increased safety and comfort concerns and, given that the safe/comfortable distance between vehicles tends to increase at a rate that is the square of the vehicles’ speed (so, going twice as fast can result in moving half as many people). Speed also tends to reduce fuel-efficiency, given air resistance and surface friction. Finally, speed tends to reduce accessibility: since more accelerating and decelerating is needed with more speed, the number of entrance and exit points to the tunnel may decrease. (Roads, for example, have many more access points than highways).
But what if we do away with the assumptions that human travellers need speed, and that human travellers are unwilling to lie on their backs within a narrow tunnel? What if, like Dracula in hibernacula, people could sleep while travelling at very slow, steady speeds in a comfy capsule capable of using narrow tunnels? If by travelling slowly the vehicle could avoid decelerating and accelerating, then the human within it would not (in theory) even be able to know that he or she was moving, and so might be able to get a very good night’s sleep. At a speed of 100 km per hour, the passenger could travel 800 km in 8 hours, to cross a natural barrier via a traffic-free, shortcut route, then wake up the next morning at their urban destination. During the day, assuming that passengers would not want to spend more than an hour or two lying down in a capsule, the system could then transport more cargo long distances, and people shorter distances, crossing under natural barriers while taking a power nap.