At Factual we are fond of raising complex questions on the future of mobility, and so we fearlessly provide expert insight and tools for thorough discussion and analysis, together with our customers and partners, to imagine feasible future scenarios, early identify challenges, shed some light upon what is to come, and ultimately be better equipped to design the mobility we want, or better put, the mobility our cities need and deserve.
That is why we have liaised with Bern Grush, a renowned expert in the field of AVs, who will join us for an exciting 3-day executive course, 4-5-6 of February in Barcelona. The headline is: “The End of Driving: Planning for Autonomous Mobility“. Check the agenda and side-event programme, and consider registering for the last available seats!
In the meanwhile, as a warm up exercise getting ready for the course, we have shared some hot questions from registered participants with Bern. Here is the outcome of our interview as a snapshot on the course content:
Factual Consulting (FC): When do you think AV technology will be ready for public use?
Bern Grush: Well, I’d like to answer that only considering purely driverless AVs – vehicles that do not need a driver and do not have a normal steering wheel, except maybe tucked away or on a tablet for emergency use.
I am very doubtful this level of automation will happen in the next few years, and I am hardly alone in this thinking. From a pure vehicle-technology perspective, we are still seeing that automatic braking is not 100% reliable — certainly not in most of the vehicles tested. What can be more important? Your foremost job as a driver is to not hit anything. That’s a basic minimum we expect from these machines.
Of course, there are still other things we think of as technical such as operating in almost any weather conditions or driving on any passable road, and we have not solved all of those yet either.
Governments are unlikely to admit autonomous fleets into general service until there is clear evidence that those vehicles are safer than human drivers. And to date there are still far too many edge cases to show they are safer.
But to acknowledge all the progress that has been made, the technology is ready for tightly controlled environments. Unfortunately, we have very few such environments. We need roadways in reasonably good condition, with unambiguous lane markings, mild-enough weather, and well-mapped with high-definition mapping. We need ways to monitor and assist driverless vehicles for failure, and places for them to pull off the road in the event of breakdown.
From here, it gets worse, because there are so many other dependencies beyond the ones I just mentioned. We also need sufficient telecommunications systems, well-distributed energy delivery, and emergency recovery. We will need the equivalent of an air-traffic control system to handle dropoff areas for when the number of dropoff requests exceeds the number of parking bays. How will it work when a significant number of vehicles in a particular area are robotaxis loading and unloading passengers with growing goods delivery fleets competing for that same curb space?
There are many other issues. Each is not hard to think about. But what is hard is how to resolve these all at once. We need to put together sufficient fleet capabilities, sufficient operating management, and with sufficient availability and reliability for a significant number of people to rely on these services so as to make it financially viable to operate such fleets. Just look at the continuing concern surrounding the ride-hailing business model to see how precarious that will be. The jump from technology trials to urban- scaled fleets is enormous.
The common assumption that “taking the driver out” will dramatically lower the cost of hailing a taxi and therefore “everyone” will switch from owning cars to hailing rides has no reliable evidence. The greatest mistake we make is to think that mostly what a service driver does is wiggle a steering wheel. A driver is making many other decisions especially at each end of a journey regarding the disposition of the vehicle. It is the “whole-system” decisions that are hard to automate. Steering is not the challenge.
Because we need a complete solution that satisfies enough users over a large region to build and operate a reliable and profitable robotaxi fleet, the question of “ready for public use” can only be: “More slowly than we are often led to expect, and more slowly than many of us wish.”
FC: When they are ready, will these only be available as taxis, or will I be able to buy one for my family or personal use?
BG: Ooph! Who would want a vehicle that can’t go on any road, in any weather, and at any time? You would need to live in a fair-weather city that is rather advanced with its AV support systems, or be satisfied using your AV only under whatever the current geofencing and weather constraints happen to be at the moment. Will it be OK if it’s snowing and the geofence shrinks for six hours to allow the city to clear the snow?
If this was your “around-town” car for shopping, or taking your kids to school, or other local trips, then sure, that might make sense for a few families as a second car. I call this inability to go anywhere no matter what “access anxiety,” and that means you would not likely purchase such a limited vehicle. Instead, if you still want to own a car, you would buy a self-driving, Level 3 vehicle – one that you can turn off and use the steering wheel. So, where does that really get us?
Rather than buy a driverless car with such limited access, most people would just rely on robotaxis or would purchase a merely self-driving car. In the early stages of AVs, you will likely see both —just as many families today might use transit to go to work or downtown but a family car to go shopping or out of town. It will be decades until a majority of people would purchase a driverless car. That will only happen when driverless goes “almost everywhere” as our Level 1 cars do today. In other words, the driverless geofence we then live in would be so extensive that we’d perceive it as “everywhere” for our purposes.
But consider that “range anxiety” remains a stubborn barrier to EVs for many people even when battery range exceeds fossil fuel range. We are not all so quick to adapt. The lingering effect of “access anxiety” will behave the same way ensuring that many people will expect the option to drive for some time, hence many will not buy driverless cars until long after they reach the Level 5 stage which looks increasingly like 2060 or 2075.
So, at first, its carefully managed robotaxis — or Level 2 and 3 family cars. And you will not be able to send a Level 3 car out on its own to run an errand or pick up your kid from soccer. You will have services for that anyway.
FC: What steps should a city or town take to ensure AVs are used safely?
BG: This is a wickedly-difficult question. Safety means a lot of things. Consider its first meaning — that AVs should not cause crashes. We expect national, provincial, or state departments of transportation to provide guidelines that constrain the vehicle registration process to disallow unsafe vehicles. That would mean that AVs being used in cities — private or taxi — would be safe without cities or towns doing anything. But is that enough? Does “safety” not also imply protection from any form of tampering or hacking that would enable a crash, a robbery, a rape, or a kidnapping? How could a city — or any government — guarantee that?
Likely cities can’t do much more than set an expectation of a very low probability of such transgressions — or to demand significant insurance against those outcomes. If a city were to force vehicle manufacturers and fleet operators to insure against all of these problems, then perhaps we could take some comfort. After all, I still use a credit card even though my credit card number has been fraudulently used three or four times. I do that because each time the card company has taken the hit.
Still there are other problems, such as unintended consequences for health, climate, growing congestion, and sprawl. These may not be considered safety issues, but they are arguably sources of great harm to human and urban well-being that cities would do well to consider.
Why I consider this a wicked-problem is that solving everything will increase costs and increasing cost draws out the time to the solutions we seek, while introducing new social-equity issues.
FC: If driverless taxis become popular, won’t that create conflict in busy places for people to be picked up and dropped off? How will that be handled?
BG: It certainly would. In fact, many cities already have problems with too many taxis or too many ride-hailing vehicles loading and unloading on our busiest streets. Don’t forget goods deliveries compete for this same space.
We will need some form of real-time operational management; a way to reserve spots to drop off and pickup. That will need in-motion queuing, realtime communications, all the kind of things that make air-traffic control work. This will have to be standardized and regulated at the national and international level. The International Standards Organization has just started a project to work on this.
FC: We should be using AV taxis for elderly and disabled travelers. Will we? Will this be safe for them?
BG: Another tough question. How elderly? How disabled? Will these users need assistance to get in an out of vehicles? They often do now. Will they be able to put on seat belts? Be nimble enough to provide smart-device commands or push a start or emergency button? Be competent to issue voice commands? We still need to provide a useable user interface and human assistance for special-needs passengers.
But for some significant portion of senior and disabled people getting to the curb and getting into and out of a vehicle unaided, and commanding a machine using a smart device is clearly possible. Will these users feel safe? Will AVs be programmed to provide time enough if these users move slowly? Carry shopping bags? Fold and stow a mobility device? Will user commands be correctly understood? What about passengers with auditory, visual or speech problems?
I think that there is both great potential and a large set of critical design problems that we have not begun to address. This is the sort of work that can start now, long before AVs are in pervasive service. In a way, we should be grateful that the roll-out of AVs will be much slower than advertised. We are far from ready for them — which is actually the major reason they will take so long. This is a larger problem than the current limitations in AI.
We who are waiting for AVs see them as not quite ready yet. While that’s true, the irony is that it is we who are the least ready. We should attend to our road, transit, curb and urban traffic matters before expecting AVs to address these for us. If we automate what we have now, we will regret the outcome.
FC: Will AV taxis disrupt, upgrade, or complement our bus systems?
BG: his can easily go either way. Cities with poor bus service that have lots of transit deserts and insufficient off-hour services could be disrupted relatively quickly. This might mean such cities retire parts of their bus services. Such a down-spiral usually has serious social equity issues, unless any new service is affordable and available to all who had depended on the original bus service. At the other extreme, these same cities could deploy new forms of transit services taking advantage of driverless micro-transit and on-demand robotaxis for example. But this is non-trivial as well. There will be long transition periods that include service changes, service disruptions, new services – perhaps trial and error – that may or may not experience the expected demand, or provide the intended improvement.
So, the change may be positive or it may leave some people worse off. What is critical for any city to start now is to ask what they want to see happen and to plan the kinds of changes and programs that are suitable for their city. The City of Toronto has started that sort of thinking and is a good example to study. The hard part about this is we do not yet know how soon and how reliable and in exactly what formats AV technology for public transit applications will take. So far, we really only see examples of robotaxis and robo-shuttles on test circuits. How will those test vehicles scale to massive fleets? To changes in the way people are picked up and dropped off? How will these fleets be managed? Owned? Priced? Subsidized?
I think the hardest part of all will be half-way through. When half of us are using old transit systems and half of us have switched. It is this half-way point that no one is looking at squarely.
FC: As you probably know, cities like Barcelona have ambitions to reduce urban space for vehicles with the Super Block concept. Can we speed up or expand this strategy with the appropriate use of AVs?
BG: I think so. And I think we should. One approach would be to begin the process of excluding private vehicles (with special exceptions) from the periphery of superblocks. As the number of superblocks grows, one should be able to find electric AV shuttle and taxi services within half the length of any superblock to travel to any other superblock, to a transit hub, or to peripheral parking areas for renting a vehicle for travel away from these areas. This way we could gradually expand the effective areas of superblocks and, of course, areas free of car ownership but still with plenty of motorized (electric) mobility for trips beyond the active transportation range.
Our book, The End of Driving, has two chapters about this concept, which we call Transit Leap. This shows the steps that can be used over three decades to turn entire regions such as the Barcelona Metropolitan Area into a region that is virtually free of private vehicles, while still providing safe, reliable mobility for trips beyond the range of walking and biking or beyond the ability of seniors, disabled or load-carrying travelers to use active transportation.
FC: How can we be sure whether AVs will mean fewer or more cars on our streets?
BG: We can’t. We have to intentionally prepare for the outcome we want. Otherwise we will get more cars. Humans overconsume whatever is cheaper, better, sweeter, or more convenient.
Consider that even in the extreme circumstance of zero private vehicle ownership, if everyone were to take on average the same number of trips for the same distances with the same vehicle sizes and the same vehicle occupancies, traffic would be the same. In reality, any city that keeps those averages but densifies would naturally see even more traffic. The only difference would be less land used for parking, which helps densify and therefore we’d tend to see still more traffic if nothing else changes. If we address only the parking issue, but ignore the trip behaviours, AVs will make traffic worse.
In reality, to reduce traffic, we need to see a combination of fewer motorized trips, more active trips, shorter trips, more use of transit, and higher occupancy when we take taxi trips. None of this is directly about automation. Automation is only an enabler — a means to an end. If we focus only on automation, as so many are, we will miss, again. The real requirement is fewer, shorter and higher occupancy automobile trips. We should be asking how to aim ourselves in that direction now. We should not wait for AVs, rather we should use AVs to help us after we are already headed in the desired direction.
The rule is: decide what you want, start preparing for that and use AVs to accelerate your solution when they are ready. What you do not want is to just wait to see what AVs will bring you. Very lazy. Very dangerous.
FC: How will it work out if private company AVs compete with our municipal bus system(s)?
BG: In a word? Badly, for your bus system. How this works out for your city depends on a few things. In small cities or in suburbs, most people who do not own a car might be better off — if they can afford the fare for robotaxis. Those who do not own a car and cannot afford the robotaxi fare would be worse off if the bus system shuts down or reduces its service offering.
In larger cities with mature, productive bus systems significant AV fleets would likely exacerbate congestion unless those fleets were high-occupancy fleets operated in very intelligent, coordinated ways. Certainly, trading a bus that carries 50 people in peak traffic for 35 robotaxi sedans or 15 robotaxi shuttles in that same traffic would undoubtedly increase congestion no matter how well behaved those taxis were.
If the municipal bus we now know cannot survive, then in addition to all the other things I mentioned earlier such as shorter, fewer, and more active trips, we must find a way to shape AV technology into an optimal substitute. Massive robotaxi fleets on their own are not going to save us.
Come join us for “The End of Driving: Planning for Autonomous Mobility”, 4-5-6 February in the beautiful city of Barcelona. Few seats are still available, so do not hesitate and write us an email at email@example.com or leave us a comment.