4 Comments
User's avatar
A B's avatar

My prediction is that robotaxis will continue to expand, but their prices will be similar to human-driven taxis, which is too expensive to replace personal car ownership in any significant numbers.

What I think is more likely to happen will be the pessimistic view, where everyone continues to own their own cars, but the cars start getting autonomous driving capability. Thus, a parent taking a kid to school no longer has to physically ride with the kid in the car, and anyone driving somewhere where it is expensive to car can just have the car drop them off and drive back empty to the driveway back home. The convenience of being able to do this sounds great at first glance, but in aggregate, it will lead to a lot of extra cars on the road, many of which having zero people on board. From a transit perspective. It may eventually force cities to start embracing congestion pricing to avoid total gridlock. And, of course, dedicated lanes will become even more important for transit to be a viable alternative.

(And, no, mass autonomous vehicles will not magically solve traffic congestion, as the space taken up by a car on the road does not magically get smaller as a result of automation, and whatever traffic flow improvements are realized by robots merging more effectively than humans will be quickly offset by ever more cars on the road).

Expand full comment
Ilya Petoushkoff's avatar

Waymo is operating commercially, so I don't think I see a lot of reasons for not getting first major commercial deployment of driverless buses somewhere very, very soon.

Expand full comment
A B's avatar

I can think of quite a few reasons why not:

1) Transit agencies, by nature, need to be very risk-averse. If Waymo encounters a problem with it's technology, they can suspend service, and people can get around some other means until the problem is fixed. But, if a transit agency tried to do that, lots of people would be left stranded.

2) Because buses are much bigger vehicles than cars, with many more people on board, the consequences of any accident are more serious (e.g. an accident caused by a software glitch can now easily kill dozens of people, rather than just two or three).

3) People expect their bus system to be functional during weather conditions that Waymo does not operate under today (think snow, ice, heavy fog, etc.).

4) If a transit agency needs to switch back to human drivers on short notice due to technical issues, it's not possible to get enough drivers quickly enough unless such drivers are already on payroll, which eliminates the labor savings of going autonomous in the first place.

5) Bus drivers are expected to do other things besides just drive the bus, that autonomous vehicle technology can't easily replicate, for example, be a human face that people can talk to who don't know where they're going, strapping and unstrapping wheelchair passengers, or acting as a security presence to deter bad behavior.

6) The wheelchair case might make autonomous buses outright illegal under ADA, unless somebody comes up with a way to allow wheelchair passengers to board a bus unassisted (which works just fine with rail, but I'm not sure it's possible on buses that load or unload passengers at regular street curbs; if it were easy, it would have been done already).

7) Transit bus drivers are generally unionized, and such unions would fight tooth and nail against any plan to replace their jobs with robots. As leverage, they could threaten that if some bus routes go autonomous that all of the bus drivers would go on strike, shutting down the entire system. Or, they could get the mechanics who maintain the buses to also go on strike in solidarity. They could also use their money to back politicians who oppose bus automation in races for positions that control the bus system.

All that said, if a transit agency were to embrace automation, the place to start would probably be rail, rather than buses. Because trains run on tracks with fixed routes, the technological problem is much simpler. The wheelchair/ADA issues go away, and people are already used to train drivers being in an enclosed cabin, away from the passengers. Some trains with grade-separated systems are, in fact, already automated, and it wouldn't be that much of a leap to automate light rail on surface streets as well, you just need to be able to trust the systems to prevent the train from running over a car or pedestrian illegally crossing the tracks.

Expand full comment
Ilya Petoushkoff's avatar

(1) There's no need for overnight network-wide automation, as can be seen from some rail systems that opted towards going driverless. Phasing allows mitigating these risks.

(2) The question is whether automation is more reliable than the driver. As soon as the answer is yes (and with Waymo, under a restricted area of service, it seems to be the case), this is in principle solved.

(3) This can be definitive to the areas where automation is more likely to happen first, and in some areas it may indeed be extremely challenging to automate because of harsh weather conditions being a recurring norm.

(4) see (1) and (2)

(5) This would require quite a bit of change in customer-agency interactions and some of the advantages of human touch will get lost on the way. This being said, it has been solved in all automated rail public transport systems around the world, and automated buses won't be all that different in that regard. There will be some customer-facing staff available where this is absolutely necessary (such as key terminals / interchanges), as well as customer assistance crews moving around the network (including ad hoc assistance to travellers who might need it). A driverless system wouldn't go anywhere near becoming completely human-less, but humans won't be driving.

(6) Kerb access is a tricky matter and this would indeed require quite a bit of thinking, including street design. I can see many more median-running bus lanes being deployed as one of the possible solutions to this.

(7) Hiring freezes and voluntary departure packages are possible approaches to gradually reduce headcount where necessary. (New customer-facing roles would require humans, too.)

However, the deployment of driverless transportation takes a lot (a lot) of time and effort, and given the level of complexity involved, this is extremely unlikely to get any faster even with the best of technology, so I wouldn't expect too much trouble with this.

---

Driverless light rail may come in sooner than driverless buses—I agree.

Expand full comment