Future of Transportation

March 6, 2020 8:50 AM

Future of Transportation Tony Seba Keynote at the 2020 N.C. DOT Transportation Summit.

Some timecode links to the key sections of this 1 hour talk:

Cars replace horses in 1913

Forecasting Technology

Technology cost curves, as production expands, the costs of technologies come down.

Energy from Solar Photovoltaics (PV) predictions see also Wikipedia stats

New Technologies are not adopted linearly but are always adopted in an S curve manner.

Smart phones, Uber and Lyft

Decline in car ownership

Battery costs and energy market share

Disruption by Electric vehicles - gas savings, life span and maintenance savings.

The tipping point for electric vehicles = 2021

Amazon investing in electric vans

Investment in autonomous vehicles

Are autonomous vehicles safe? Insurance costs for human drivers will go up.

Last mile, remote piloting

Transport as a service (fleets) monthly payments will be cheaper than owning.

Implications: financial, social, environmental

Effect on oil demand - future exploration may be uncommercial

Conclusion

posted by Lanark (20 comments total)

Relevant to my interests and work.

Per Tony Seba's website, he's an author, thought leader and Silicon Valley (SV) entrepreneur. His blog is out of date (last updated in 2018) but has interesting ideas (focus on zero emission miles, not zero emission vehicles -- how many carbon-free miles were traveled, instead of how many vehicles sold).

He's also part of RethinkX Transportation, which foresees that by 2030, "95 percent of U.S. passenger miles traveled will be served by on-demand autonomous electric vehicles owned by fleets, not individuals, in a new business model we call “transport-as-a-service” (TaaS)."

I think it's not going to happen that soon, but every year, there are more "driver augmentation" features rolled out in standard vehicles, making the transition to automation progression up an incline instead of jumping whole steps, so I could well be proven wrong.

[And I just starting reading a research proposal that states "It is predicted that half of the vehicles sold and up to 40% of vehicle travel could be autonomous in the 2040s."]

Given the number of fatalities and serious injuries in the US every day due to vehicle crashes, I still thing autonomous vehicles provide a potential for safer travel. That said, we don't have to rely on automation to make us safer. Oslo had one traffic death in 2019, which occurred when a driver ran his car into a fence this summer — a decline from five traffic fatalities in 2018. But this came in part by taking back roads from cars and giving them back to pedestrians and cyclists:

The city removed more than 1,000 parking spaces, added more bike lanes and sidewalks, improved bike-share, launched public transit improvements and banned cars in some downtown areas.

As for EVs, and truly zero emission vehicles (EVs charged from renewable resources), I'll note that with the robust growth in photovoltaics, we're still only up to PV deployments sufficient to supply about 3% of global electricity demand (2019 numbers). I haven't looked for other nations or regions aiming to operate zero-emission electrical systems, but PNM, New Mexico's major energy holding company, aims to be emissions-free by 2040.

My utopian transportation dream is not 95% automation and TaaS as envisioned by SV disruptors, but 95% "alternative" transportation use -- public transit, bicycling, and walking, with the support necessary for those who have limited personal mobility to be able to operate easily in that system.
posted by filthy light thief at 9:26 AM on March 6, 2020 [3 favorites]

One point: I think there's a scenario where the driver augmentation features reduce accidents to the point that they make moot the very expensive, difficult*, and culturally conflicted proposition of non-ownership.

A relatively cheap Subaru with EyeSight today makes the most common accidents difficult to perpetrate; i.e. rear-end collisions.
Insurance companies looking at accident incidence may not see a difference between 2030 human and robot-driven automobiles.


*Caveat: I don't believe in the tooth fairy or computer software replacing human drivers in my lifetime, appealing as both ideas are.
posted by backsaw at 10:16 AM on March 6, 2020 [1 favorite]

Too bad the DOT is facing a major budget crunch because they do some of the coolest work in government
posted by shaademaan at 10:17 AM on March 6, 2020

Carmakers and technology companies have concluded that making autonomous vehicles is going to be harder, slower and costlier than they thought
...Mr. Salesky said Argo and many competitors had developed about 80% of the technology needed to put self-driving cars into routine use — the radar, cameras and other sensors that can identify objects far down roads and highways. But the remaining 20%, including developing software that can reliably anticipate what other drivers, pedestrians and cyclists are going to do, will be much more difficult.

A car which can only self-drive 80 or 90% of the time is more dangerous than a car which can't self drive at all - they just increase driver distraction.

One way to get that figure up to 100% would be to completely ban human drivers and pedestrians from certain roads allowing the more predictable robots to take over. You could also vastly increase the speed limits to match the capabilities of fast driving computers.
posted by Lanark at 10:37 AM on March 6, 2020 [1 favorite]

Relevant to my interests and work.

Same for me - except my work is in computer security. I co-founded a company that (pen)tests systems. And here's the thing: EVERY single system we test, we find flaws. I'm driving an old Aygo and I will soon have to replace it. I will have trouble finding a good replacement since every new car these days is connected to the internet. I know the sorry state of software security, and I don't WANT my car connected to the internet in any way. At around 39 minutes into the presentation, he predicts at most two systems of autonomous driving will "survive" and dominate the market. And the examples he gives: iOS/Android, and Mac/Windows. Now go check malware numbers for those platforms. And then imagine some pimple faced teenager in some suburb somewhere launching botnet script that takes out a massive number of those internet connected cars for shits and giggles and bragging rights at school.

No thanks.
posted by DreamerFi at 10:52 AM on March 6, 2020 [6 favorites]

(and that's before even thinking of whole new classes of attacks on vehicle autonomy. And no, this is not a NEW attack class)
posted by DreamerFi at 10:55 AM on March 6, 2020 [2 favorites]

One way to get that figure up to 100% would be to completely ban human drivers and pedestrians from certain roads allowing the more predictable robots to take over. You could also vastly increase the speed limits to match the capabilities of fast driving computers.

Roads are designed fundamentally more than anything else around the stopping sight distance calculation, which is a very basic physics equation: SSD = speed * perception/reaction time + speed^2 / deceleration rate. The first half is how far you travel before you apply the brakes, the second half is how quickly you stop once you hit the brakes. This one calculation (or a slight variation of it) underlies so much of highway engineering; how far back you need obstacles from intersections, how long a yellow light should be, how to stripe a passing lane, how sharp a curve or hill can be.

Deceleration rate won't change with autonomous vehicles; just as there are old cars with shitty brakes on rainy days now, there will eventually in the future be old robot cars with shitty brakes on rainy days. A road that is currently designed for 90 km/h (about 55 mph) has an SSD of 160m (rounded up from 155.5), based on AASHTO Green Book calculations, which assume 2.5 seconds of perception/reaction time. Let's be as generous as possible, and assume that robots can detect any problem instantaneously. They can't - if a rock falls on a road or a tire blows out on the car in front of you it will take a fraction of a second for it to be evident to an observer that the rock won't roll out of the lane -- but I'm feeling generous.

So we can substitute the 2.5 seconds of perception/reaction time with 0 and do the math; it turns out that even with impossible speed in perceiving and reacting to problems, to stop in 160m, you need to be travelling at a max speed of 115 km/h, (about 70 mph). Which is about 25 km/h or 15 mph faster than the current speed the road is designed for. (If you assume half a second to respond, which is more realistic, then you can only get a 20 km/h increase.) This is not a vast increase; if you look at how fast drivers are usually going on a freeway like that, it's barely an increase at all. Which is the terrifying part -- drivers today are already driving more or less as if they were perfect robots.
posted by Homeboy Trouble at 11:07 AM on March 6, 2020 [4 favorites]

Maybe that means the driving standards and standard equations are wrong?
posted by The_Vegetables at 11:17 AM on March 6, 2020

Maybe that means the driving standards and standard equations are wrong?
By necessity they are based on the worst road conditions and a vehicle with marginal brakes, a high performance vehicle in perfect weather is always going to be capable of exceeding the limit without crashing.
posted by Lanark at 11:26 AM on March 6, 2020 [2 favorites]

Yeah, the problem with human drivers is that they over-drive their perceptions, so if you improve sight-lines on roads, drivers just drive faster. If you want to make a road safer, you should make it feel utterly terrifying for the human driver. But if you do that, you get people complaining that they can't see around that corner and the trees should be trimmed...

The advantage of self-driving vehicles is gonna be a very clear liability chain back to deep pockets, which means we might actually have a prayer of restraining vehicles to safe speeds.
posted by straw at 11:28 AM on March 6, 2020 [2 favorites]

The future of transportation is (trains + light rail + busses + bicycles) - cars.

(autocarrot thinks busses is spelled with two s's but it looks wrong to me)
posted by antinomia at 11:51 AM on March 6, 2020 [5 favorites]

And for door-to-door TOP speed doesn't have to increase a lot at all. Average speed going up because of less traffic, and less traffic disruptions (such as traffic lights) is a far bigger time-win.
posted by DreamerFi at 11:52 AM on March 6, 2020 [1 favorite]

The advantage of self-driving vehicles is gonna be a very clear liability chain back to deep pockets, which means we might actually have a prayer of restraining vehicles to safe speeds.

The existing clear liability chain to deep pockets has had negligible restraining effects on what the likes of Uber, Tesla and Google do today.
posted by Homeboy Trouble at 11:57 AM on March 6, 2020 [2 favorites]

The part of this future that still seems doubtful to me is that if Autopilot never gets to 100% coverage then you will still need a driver for that Uber/Lyft, and then the cost of paying the driver is going to wipe out any savings from cheaper energy. So you still end up with people driving their own cars but it will be cheaper so there will be more of them and more congestion.

Also in most countries Petrol and Diesel are heavily taxed, if that tax revenue disappears governments will just have to find it somewhere else.
posted by Lanark at 12:20 PM on March 6, 2020

If you want to make a road safer, you should make it feel utterly terrifying for the human driver
This would work for awhile, but people adjust to being terrified. California highway 17 between Los Gatos and Santa Cruz is a good example. It's terrible in every way, but people routinely drive 15-20 mph over, and there are 1-5 crashes a day in bad weather
posted by Maxwell's demon at 12:38 PM on March 6, 2020 [4 favorites]

That just looks like a normal mountain highway to me, and you can see the changes to straighten curves, add lanes, and remove vegetation from the section of "Old Santa Cruz Highway" that runs parallel.
posted by The_Vegetables at 1:41 PM on March 6, 2020

It is early days. I think there's a lot to be excited about with EVs, but there's a lot of uncertainty about how the major obstacles will be overcome. I'm not a believer in the Gospel of the market/tech/engineers inevitably coming through.

For one thing, there's not enough Class 1 nickel to grow the EV market to 2x what it now with long-range models. Class 1 nickel is absolutely irreplaceable for the Generation 3b cathodes that are best of class right now. Attempted mining and metallurgical solutions are proving to be very expensive.
There are cheaper cathode chemistries - based on abundant minerals - that are slightly more modest in power and range, and safer too, but that's so unsexy.

Charging infrastructure becomes much hairier as we get more adoption. Just the volume of copper required is a bit of a head scratcher from a supply perspective.

It is damned interesting, I'll grant you that.
posted by Glomar response at 1:44 PM on March 6, 2020 [3 favorites]

I looked at the clip with "financial, social, environmental" effects.

I'm an urban planner, not an economist, and think it is essential to point out the financial burden of private car ownership, BUT

I am HUGELY skeptical of his claims that switching to Transportation-as-a-Service (TaaS) will result in
- $1 trillion back in the pockets of U.S. residents (is TaaS somehow free? without any costs anywhere?)
- $1 trillion productivity-->income boost because TaaS gives us more time to do stuff besides driving (how has increased productivity and a burgeoning GDP worked out for the average U.S. worker.?)
posted by spamandkimchi at 3:45 PM on March 6, 2020 [2 favorites]

We already know how to make roads safer.

It's merely a matter of political will to build them that way.
posted by rum-soaked space hobo at 10:46 PM on March 6, 2020 [2 favorites]

A Trip Through New York City in 1911: still lots of horses! and trolleys :P (via)

re: disruption, The End of Pay-TV could be a useful analogy in terms of (economic) ecosystem collapse:

The macro-point here is that the traditional TV players haven’t just embraced D2C, they’ve decided they’re done trying to stem the decline of Pay-TV. They are reducing their investments in the channel, pulling out many of the investments they’ve already started, and begun deliberately speeding the collapse in the hopes that if they burn their boats relatively faster or harder than their peers, they’ll be best positioned in the future. This doesn’t mean they don’t still want to pull cash out of the old system — they do — but they’re not optimizing for it, even when the ROIC is clear. And they’re right to do so.

The knock on-effects here are profound... most important is the consumer side. From 2010-2019, the perceived value of Pay-TV eroded because (1) better, lower cost substitutes emerged, and (2) Pay-TV pricing continued to increase. even though the quality and volume of content available in the Pay-TV bundle grew. What’s set to happen from 2020 onward is quite different. Pay-TV will, for the first time, get rapidly worse on an absolute basis as it is undermined, underfunded, and eventually defunded.

What’s more, most of what’s available in the Pay-TV ecosystem will be available elsewhere via better experiences and at lower prices (and often earlier, too)... This means that Pay-TV decline estimates are wrong. It’s not about escalation or curve modelling. The floor will suddenly start falling out. This is almost impossible to model correctly; there’s probably no equivalent example in media where all partners not only rapidly change distribution channel and halt investments, but also try to pivot from B2B to B2C monetization.

like if (when?) automakers suddenly pivot to autonomous fleet vehicles (as a service ;) because it is so much more practical/economical of course -- when run on the sun!
posted by kliuless at 1:49 AM on March 7, 2020 [1 favorite]