One of us recently attended an Automated Vehicle conference and compiled these comments. Because we are fairly critical of much current AV-thinking, we decided against disclosing which conference or which speakers/attendees we describe. We thought that prudent and fair, even if lacking in courage. No one yet understands how the autonomous vehicle future will roll out in any detail, and most people thinking about this usually approach prognostication either from the perspective of what they would like to see, or perhaps the one they fear the most.

The mixture of utopian and dystopian in a single conference was jarring. We suspect that this is the current state in most such conferences and given Gartner’s Hype Cycle view, this will continue well into the early 2020’s.

Conference Notes

A key operating assumption held by many at the conference was: “AVs will be here in 5 years.” This is because several auto brands have promised this, as has Google (until they recently backed away from the promise of full-autonomy everywhere.) At the same time, the term “AV” is assumed by many to mean “no driver, no parking, few crashes, mostly Uber-robotaxis, etc.)

[Our view: what is really being promised “by 2020” is a car that can be manually switched in/out of full autonomy mode for long highway and rural stretches, but will require a licensed driver and as many parking spots as now. In a recent, independent Car&Driver test of four, 2016 models (BMW, Infiniti, Mercedes, Tesla), these vehicles currently need operator corrections from once every km to once every four km over a mixture of highway, rural and city driving.]

For some speakers, this meant SAE L5 (Level 5=no driver needed) vehicles would “be in showrooms” and available to household users around 2020. The assumption that seems to be at play is that the gap between Advanced Driver Assistance Systems (ADAS, or SAE L2, L3) described above and full autonomy (L5) is just one more R+D increment over the next lower, L4. In other words, “we are almost there, and the technology is on an exponential curve anyway,” so “get ready”.

For other speakers, the mismatch between technology on one hand and policy, infrastructure, urban planning, social acceptance, special driving conditions, etc. on the other hand seems to create a daunting (and confusing) barrier to AVs being useable from a personal household perspective in the 2020s. This remained unresolved at the conference.

[Our view: an L5 vehicle usable anywhere that current vehicles can drive now is closer to 2040. What will be available in 2020 will be remarkable but will not end the employment needed to drive the vehicles the world will continue to require.]

Buses will need drivers for a long time and stewards instead of drivers for a long time thereafter. Heavy goods vehicles in platoons are expected to need a human steward even if driving is unnecessary. Hence, the “no-driver” scenarios for the type of fleet-purposes we have now are much further away than 2020, although a majority appears to believe this will indeed take place at some point.

[Our view: an L5 vehicle that can go “most” places, but not all, will induce “access anxiety” a disorder worse than the “range anxiety” that dampens the sale of electric vehicles. Few would buy a vehicle that only goes “most” places.]

A smaller, but significant number saw that autonomous, six-passenger minibuses are deployable now (for slow, constrained, robotic, fixed, first-last mile/campus operation), but significant speaker attention was directed toward generalized household and robo-taxi formats.

[Our view: this reflects our society’s preferences for personal vehicles (household car) or personal ride (robotaxi), and our society’s always-minimal personal preference for transit. The AV transit that is already available in the EU has the status appeal of a parking shuttle in North America.]

This led to a concern for what to do about a possible new flood of vehicles, and a demand for more VKT, more parking, and more lanes or some combination of these outcomes.

[Our view: These AV conference speakers seemed to be more thoughtful about the utopia/dystopia equation than we experienced at such conferences in 2015. As Gartner predicted, the positive hype is dampening and starting to turn negative.]

Overall, the speakers and audience focused more time on primary effects (safety, jobs, congestion, sprawl, parking, efficiency, insurance) and less time grappling with secondary effects (retraining, distracted driving with L2 & L3, transit, shifts in land value), although privacy, legal, standards and regulation were described as important and unresolved. One person worried whether we would replace the street parking areas with massive drop-zones (not even a kiss from the driver!), and another whether we’ll need to replace the landscaping around buildings with asphalt to accommodate this.

No one mentioned the effect on transit hubs (i.e., would they grow, shrink or become smaller or more numerous), although such hubs appeared to be assumed by those mentioning transit. No one discussed the threat to transit-oriented development, potentially engendered by replacing larger hubs with more numerous smaller hubs.

[Our view: we are reminded of the early 20th century tendency to design cities around the automobile. Many of the questions and concerns betray an assumption that we could redesign our cities around the AV. When you picture AVs as four-person sedans (which is what all the test vehicles look like on YouTube), it is hard to break out of a 100-year-old urban design paradigm. Unfortunately, we could easily repeat (worsen!) our most egregious urban error of the past century. Fortunately, some others also notice this.]

A portion saw public transit on roads somewhat or completely replaced by autonomous transit. Rail will survive. Replacement of bus routes at the edges was suggested as likely good since the edges are usually poorly served and highly subsidized. A more significant group of speakers did not even consider transit.

[Our view: this is the usual mix of the beliefs: “transit would stay the same” and “transit would disappear”. We have seen both arguments many times. But if these are our only two choices, the latter will win and equity would be threatened, perhaps severely.]

Private robotaxis were seen as obvious by most; many seemed to believe robotaxis might become dominant; one senior planner thought they would account for 100% of personal travel; another thought they would not become significant until 2045. A few assume a relatively even split between household vehicles and robotaxis.elephant3blindmen

There was little agreement what to do re the possible (a few thought probable) flood of vehicles. A lone audience member opined that during a long period of L3 vehicles (after 2020 and for 30 or 40 years) the personally owned L3 semi-autonomous vehicle would generate additional need for lanes and parking within cities—i.e., we would need more infrastructure rather than less if we just “wait-and-see”. This is because only a high robocab/high transit scenario would reduce vehicle population, and behavioral studies (mostly stated preferences or polling of experts) so far, say that vehicle use would go up (both VKT and PKT) far more than vehicle ownership would drop.

Far more complex interaction-effects were not discussed. Especially peripheral was speculation about other technologies such as batteries, drones, additive manufacturing, and virtual reality as a way to expand telework.

[Our view: We apparently have enough to think about without straying into adjacency effects. This exacerbates our inability to project accurately an agreed future.]

The final speaker pleaded that we not wait-and-see what Detroit or Silicon Valley want us to buy, but rather determine the world we want and plan communities and transportation systems to suit that world. Invent, then demand the world we want instead of waiting for it to wash over us.

[Our view: He is right. Unless we invent the future we want, we will get one very much like the one we have. Except different in ways that cannot be predicted.]

According to Gartner, we should know this elephant better after 2020. Surely by 2025…

URLs as postscripts…

Leave a Reply