Submitted by brad on Wed, 2014-07-23 15:32.
I have many more comments pending on my observations from the recent AUVSI/TRB Automated Vehicles Symposium, but for today I would like to put forward an observation I made about two broad schools of thought on the path of the technology and the timeline for adoption. I will call these the aggressive and conservative schools. The aggressive school is represented by Google, Induct (and its successors) and many academic teams, the conservative school involves car companies, most urban planners and various others.
The conservative (automotive) view sees this technology as a set of wheels that has a computer.
The aggressive (digital) school sees this as a computer that has a set of wheels.
The conservative view sees this as an automotive technology, and most of them are very used to thinking about automotive technology. For the aggressive school, where I belong, this is a computer technology, and will be developed — and change the world — at the much faster pace that computer technologies do.
Neither school is probably entirely right, of course. It won’t go as gung-ho as a smartphone, suddenly in every pocket within a few years of release, being discarded when just 2 years old even though it still performs exactly as designed. Nor will it advance at the speed of automotive technology, a world where electric cars are finally getting some traction a century after being introduced.
The conservative school embraces the 4 NHTSA Levels or 5 SAE levels of technology, and expects these levels to be a path of progress. Car companies are starting to sell “level 2” and working on “level 3” and declaring level 4 or 5 to be far in the future. Google is going directly to SAE level 4.
The two cultures do agree that the curve of deployment is not nearly-instant like a smartphone. It will take some time until robocars are a significant fraction of the cars on the road. What they disagree on is how quickly that has a big effect on society. In sessions I attended, the feeling that the early 2020s would see only a modest fraction of cars being self-driving meant to the conservatives that they would not have that much effect on the world.
In one session, it was asked how many people had cars with automatic cruise control (ACC.) Very few hands went up, and this is no surprise — the uptake of ACC is quite low, and almost all of it is part of a “technology package” on the cars that offer it. This led people to believe that if ACC, now over a decade old, could barely get deployed, we should not expect rapid deployment of more complete self-driving. And this may indeed be a warning for those selling super-cruise style products which combine ACC and lanekeeping under driver supervision, which is the level 2 most car companies are working on.
To counter this, I asked a room how many had ridden in Uber or its competitors. Almost every hand went up this time — again no surprise. In spite of the fact that Uber’s cars represent an insignificant fraction of the deployed car fleet. In the aggressive view, robocars are more a service than a product, and as we can see, a robocar-like service can start affecting everybody with very low deployment and only a limited service area.
This dichotomy is somewhat reflected in the difference between SAE’s Level 4 and NHTSA’s. SAE Level 4 means full driving (including unmanned) but in a limited service area or under other limited parameters. This is what Google has said they will make, this is what you see planned for services in campuses and retirement communities. This is where it begins, and grows one region at a time. NHTSA’s levels falsely convey the idea that you slowly move to fully automated mode and immediately do it over a wide service area. Real cars will vary as to what level of supervision they need (the levels) over different times, streets and speeds, existing at all the levels at different times.
Follow the conservative model and you can say that society will not see much change until 2030 — some even talk about 2040. I believe that is an error.
Another correlated difference of opinion lies around infrastructure. Those in the aggressive computer-based camp wish to avoid the need to change the physical infrastructure. Instead of making the roads smart, make the individual cars smart. The more automotive camp has also often spoken of physical changes as being more important, and also believes there is strong value in putting digital “vehicle to vehicle” radios in even non-robocars. The computer camp is much more fond of “virtual infrastructure” like the detailed ultra-maps used by Google and many other projects.
The two approaches will also clash when it comes to deciding how to measure the safety of the products and how they should be regulated, which will be a much larger battle. More on that later.
Submitted by brad on Mon, 2014-07-14 13:59.
It’s a big week for Robocar conferences.
In Berkeley, yesterday I attended and spoke at the “Robotics: Science and Systems” conference which had a workshop on autonomous vehicles. That runs to Wednesday, but overlapping and near SF Airport is the Automated Vehicles Symposium — a merger of the TRB (Transportation Research Board) and AUVSI conferences on the same topic. 500 are expected to attend.
Yesterday’s workshop was pretty good, with even a bit of controversy.
- Ed Olson on more of the lessons from aviation on handoff between automation and manual operation. This keeps coming up a a real barrier to some of the vehicle designs that have humans share the chores with the system.
- Jesse Levinson of Stanford’s team showed some very impressive work in automatic calibration of sensors, and even fusion of LIDAR and camera data, aligning them in real time in spite of movement and latency. This work will make sensors faster, more reliable and make fusion accurate enough to improve perception.
- David Hall, who runs Velodyne, spoke on the history of their sensors, and his plans for more. He repeated his prediction that in large quantities his sensor could cost only $300. (I’m a bit skeptical of that, but it could cost much, much less than it does today.) David made the surprising statement that he thinks we should make dedicated roads for the vehicles. (Surprising not just because I disagree, but because you could even get by without much LIDAR on such roads.)
- Marco Panove of Stanford showed research they did on Taxi models from New York and Singapore. The economics look very good. Dan Fagnant also presented related research assuming an on-demand semi shared system with pickup stations in every TAZ. It showed minimal vacant miles but also minimal successful rideshare. The former makes sense when it’s TAZ to TAZ (TAZs are around a square mile) but I would have thought there would be more rideshare. The conclusion is that VMT go up due to empty miles, but that rideshare can partially compensate, though not as much as some might hope.
- Ken Laberteaux of Toyota showed his research on the changing demographics of driving and suburbs. Conclusion: We are not moving back into the city, suburbanization is continuing. Finding good schools continues to drive people out unless they can afford private school are are childless.
The event had a 3-hour lunch break, where most went to watch some sporting event from Brazil. The Germans at the conference came back happier.
Some good technical talks presented worthwhile research
- Sheng Zhao and a team from UC Riverside showed a method to get cm accuracy in position and even in pose (orientation) from cheap GPS receivers, by using improved math on phase-matching GPS. This could also be combined with cheap IMUs. Most projects today use very expensive IMUs and GPSs, not the cheap ones you find in your cell phone. This work may lead to being able to get reliable data from low cost parts.
- Matthew Cornick and a team from Lincoln Lab at MIT showed very interesting work on using ground penetrating radar to localize. With GPR, you get a map of what’s below the road — you see rocks and material patterns down several feet. These vary enough, like the cracks and lines on a road, and so you can map them, and then find your position in that map — even if the road is covered in snow. While the radar units are today bulky, this offers the potential for operations in snow.
- A team from Toyota showed new algorithms to speed up the creation of the super-detailed maps needed for robocars. Their algorithms are good at figuring out how many lanes there are and when they start and stop. This could make it much cheaper to build the ultramaps needed in an automatic way, with less human supervision.
The legal and policy sessions got more heated.
- Bryant Walker Smith laid out some new proposals for how to regulate and govern torts about the vehicles.
- Eric Feron of Georgia Tech made proposals for how to do full software verification. Today formally proving and analysing code for correctness takes 0.6 hours per line of code — it’s not practical for the 50 million line (or more) software systems in cars today. Jonathan argues it can be made cheaper, and should be done. Note that fully half the cost of developing the 787 aircraft was software verification!
The final session, on policy included:
- Jane Lappin on how DoT is promoting research.
- Steve Shladover on how we’re all way to optimistic on timelines, and that coming up with tests to demonstrate superior safety to humans is very far away, since humans run 65,000 hours between injury accidents.
- Myself on why regulation should keep a light touch, and we should not worry too much about the Trolley Problem — which came up a couple of times.
- Raj Rajkumar of CMU on the success they have had showing the CMU/GM car to members of congress.
Now on to the AVS tomorrow.
Submitted by brad on Tue, 2014-06-24 09:45.
Some recent press and talks:
Earlier in June I sat down with “Big Think” for an interview they have titled “Robocars 101” explaining some of the issues around the cars.
I also did a short interview on NPR’s “All Things Considered” not long after Google’s new car was announced. What you might find interesting is how I did it. I was at a friend’s house in Copenhagen and went into a quiet room where they called me on my cell phone. However, I also started a simple audio recorder app on my phone. When we were done, I shared the mp3 of a better sample from the same microphone with them, which they mixed in.
As a result, the interview sounds almost like it was done in-studio instead of over an international cell phone call.
Videos of my talks at Next Berlin at at Dutch Media Future Week 2014 are also up. And a shortened talk at Ontario Centers for Excellence Discovery 2014 in Toronto May 12. There we had the Governor General of Canada as our opening act. :-) That’s just 3 of the 11 events I was at on that trip.
Completely off the Robocar track is a short interview with CNBC where I advise people to invest in Bitcoin related technology, not in bitcoins.
Submitted by brad on Sun, 2014-06-22 20:51.
So far it’s been big players like Google and car companies with plans in the self-driving space. Today, a small San Francisco start-up named Cruise, founded by Kyle Vogt (a founder of the web video site Justin.tv) announces their plans to make a retrofit kit that will adapt existing cars to do basic highway cruise, which is to say, staying in a lane and keeping pace behind other cars while under a driver’s supervision.
I’ve been following Cruise since its inception. This offering has many similarities to the plans of major car companies, but there are a few key differences:
- This is a startup, which can be more nimble than the large companies, and having no reputation to risk, can be bolder.
- They plan to make this as a retrofit kit for a moderate set of existing cars, rather than custom designing it to one car.
They’re so dedicated to the retrofit idea that the Audi A4 they are initially modifying does not even have drive-by-wire brakes like the commonly used hybrid cars. Their kit puts sensors on the roof, and puts a physical actuator on the brake and another physical actuator on the steering wheel — they don’t make use of the car’s own steering motor. They want a kit that can be applied to almost any car the market tells them to target.
They won’t do every car, though. All vendors have a strong incentive to only support cars they have given some solid testing to, so most plans don’t involve retrofit at all, and of course Google has now announced their plans to design a car from scratch. Early adopters may be keen on retrofit.
I rode in the car last week during a demo at Alemeda air station, a runway familiar to viewers of Mythbusters. There they set up a course of small orange cones, which are much easier to see than ordinary lane markings, so it’s hard to judge how well the car does on lane markings. It still has rough edges, to be sure, but they don’t plan to sell until next year. In the trial, due to insurance rules, it kept under 40mph, though it handled that speed fine, though drifted a bit in wider parts of the “lane.”
On top is an aerodynamic case around a sensor pack which is based on stereo cameras and radar from Delphi. Inside is just a single button in the center arm console to enable and disable cruise mode. You take the car to the lane and push the button.
All stuff we’ve seen before, and not as far along, but the one key difference — being a nimble startup — may make all the difference. Only early adopters will pay the $10,000 for a product where you must (at least for now) still watch the road, but that may be all that is needed.
Submitted by brad on Sun, 2014-06-22 11:30.
On my recent wanderings in Europe, I became quite enamoured by Google’s
latest revision of transit directions. Google has had transit directions for
some time, but they have recently improved them, and linked them in more cities
to live data about where transit vehicles actually are.
The result not a mere incremental improvement, it’s a game-changing increase
in the utility of decent transit. In cities like Oslo and London, the tool
gives the user the ability to move with transit better than a native. In the
past, using transit, especially buses, as a visitor has always been so frustrating
that most visitors simply don’t use it, in spite of the much lower cost compared
to taxis. Transit, especially when used by an unfamiliar visitor, is slow and
complex, with long waits, missed connection and confusion about which bus
or line to take during shorter connections, as well as how to pay.
Not so any more. With a superhuman ability, your phone directs you to transit stops
you might not figure out from a map, where the right bus usually appears quite quickly.
Transfers are chosen to be quick as well, and directions are given as to which direction to
go, naming the final destination as transit signs often do, rather than the compass direction. It’s optimized by where the vehicles actually are and predicted to be, and this
will presumably get even better.
By making transit “just work” it becomes much more useful, and gives us a taste of the
robocar taxi world. That world is even easier, of course — door to door with no
connections and no need for you to even follow directions. But while Uber also shows us
that world well in user experience, Uber is expensive, as are cabs, while transit is closer
in cost to the anticipated robocar cost of well below $1/mile.
It also helps to have transit systems with passes or contactless pay cards, to avoid the hassles of payment.
Why does this work so well? In the transit-heavy cities, it turns out there are often 2, 3 or even 4 ways to get to your destination via different transit lines and connections. The software is able to pick among them in a way even a native couldn’t, and one is often leaving soon, and it finds it for you.
In some cities, there is not live data, so it only routes based on schedules. This cuts
the utility greatly. From a user experience standpoint, it is often better to give people
a wait they expect than to do a better job but not give accurate expectations.
What’s clear now is that transit agencies should have done this a lot sooner. Back in the 1980s
a friend of mine built one of the first systems which tracked transit vehicles and gave
you a way to call to see when the bus would come, or in some cases signs on the bus stops.
Nice as those were they are nothing compared to this. There is not much in this technology
that could not have been built some time ago. In fact, it could have been built even
before the smartphone, with people calling in by voice and saying, “I am at the corner of X and
Y and I need to get to Z” with a human helper. The cost would have actually been worth it
because by making the transit more useful it gets more riders.
That might be too expensive, but all this needed was the smartphone with GPS and a
data connection, and it is good that it has come.
In spite of this praise, there is still much to do.
- Routing is very time dependent. Ask at 1:00 and you can get a very different answer than you get asking at 1:02. And a different one at 1:04. The product needs a live aspect that updates as you walk and time passes.
- The system never figures out you are already on the bus, and so always wants to route you as though you were standing on the road. Often you want to change plans or re-look up options once you are on the vehicle, and in addition, you may want to do other things on the map.
- Due to how rapidly things change, the system also needs to display when multiple options are equivalent. For example, it might say, “Go to the train platform and take the B train northbound.” Then due to how things have change, you see a C train show up — do you get on it? Instead, it should say, “Take a B, C or E train going north towards X, Y or Z, but B should come first.”
- For extra credit, this should get smarter and combine with other modes. For example, many cities have bikeshare programs that let you ride a bike from one depot to another. If the system knew about those it could offer you very interesting routings combining bikes and transit. Or if you have your own bike and transit lines allow it on, you could use that.
- Likewise, you could combine transit with cabs, getting a convenient route with low walking but with much lower cab expense.
- Finally, you could also integrate with one-way car share programs like car2go or DriveNow, allowing a trip to mix transit, car, bike and walking for smooth movement.
- Better integration with traffic is needed. If the buses are stuck in traffic, it’s time to tell you to take another method (even cycling or walking) if time is your main constraint.
- Indoor mapping is needed in stations, particularly underground ones. Transit agencies should have beacons in the stations or on the tracks so phones can figure out where they are when GPS is not around. Buses could also have beacons to tell you if you got on the right one.
- The systems should offer an alert when you are approaching your stop. Beacons could help here too. For a while the GPS map has allowed the unfamiliar transit rider to know when to get off, but this can make it even better.
- This is actually a decent application for wearables and things like Google glass, or just a bluetooth earpiece talking in your ear, watching you move through the city and the stations and telling you which way to go, and even telling you when you need to rush or relax.
- In some cities going onto the subway means loss of signal. There, storing the live model for relevant lines in a cache would let the phone still come up with pretty good estimates when offline for a few minutes.
A later stage product might let you specify a destination and a time, and then it will buzz you when it’s time to start walking, and guide you there, through a path that might include walking, bike rides, transit lines and even carshare or short cab rides for a fast, cheap trip with minimal waiting, even when the transit isn’t all that good.
Submitted by brad on Mon, 2014-06-09 19:48.
I’m in the home stretch of a long international trip — photos to follow — but I speak tomorrow at Lincoln Center on how computers (and robocars) will change the worlds of finance. In the meantime, Google’s announcement last month has driven a lot of news in the Robocar space worthy of reporting.
On the lighter side, this video from the Conan O’Brien show highlights the issues around people’s deep fear of being injured by machines. While the video is having fun, this is a real issue that will dominate the news when the first accidents and injuries happen. I cover that in detail in my article about accidents but the debate will be a major one.
Nissan announced last year that it would sell cars in 2020. Now that Tesla has said 2016, Google has said civilians will be in their small car within a year and Volvo has said the same will happen in Sweden by 2017, Nissan CEO Carlos Ghosn has said they might do it 2 years earlier.
As various locations rush to put in robocar laws, in Europe they are finally getting around to modifying the Vienna convention treaty, which required a human driver. However, the new modifications, driven by car companies, still call for a steering wheel that a driver can use to take over (as do some of the US state laws.) These preclude Google’s new design, but perhaps with a bit of advance warning, this can be fixed. Otherwise, changing it again will be harder. Perhaps the car companies — none of whom have talked about anything like Google’s car with no controls — will be happy with that.
The urban test course at the University of Michigan, announced not very long ago, is almost set to open — things are moving fast, as they will need to if Michigan is to stay in the race. Google’s new prototype, by the way, is built in Michigan. Google has not said who but common speculation names not a major car company, but one of their big suppliers.
The Ernst & Young auto research lab (in Detroit) issued a very Detroit style forecast for autonomous vehicles which said their widespread use was 2 decades away. Not too surprising for such a group. Consultants are notoriously terrible at predictions for exponential technology. Their bad smartphone predictions are legendary (and now erased, of course.) A different study predicts an $87 billion market — but the real number is much larger than that.
This article where top car designers critique Google’s car illustrates my point from last week how people with car company experience are inclined to just not get it. But at the same time some of the automotive press do get it.
Submitted by brad on Sun, 2014-06-01 05:15.
It’s not too surprising that the release of images of Google’s prototype robocar have gotten comments like this:
Revolutionary Tech in a Remarkably Lame Package from Wired
A Joy Ride in Google’s Clown Car says Re/Code
I’ve also seen comparisons to the Segway, and declarations that limited to 25 mph, this vehicle won’t get much adoption or affect the world much.
Google’s own video starts with a senior expressing that it’s “cute.”
I was not involved in the specifics of design of this vehicle, though I pushed hard as I could for something in this direction. Here’s why I think it’s the right decision.
First of all, this is a prototype. Only 100 of this design will be made, and there will be more iterations. Google is all about studying, learning and doing it again, and they can afford to. They want to know what people think of this, but are not scared if they underestimate it at first.
Secondly, this is what is known as a “Disruptive Technology.” Disruptive technologies, as described in the Silicon Valley bible “The Innovators Dilemma” are technologies that seem crazy and inferior at first. They meet a new need, not well understood by the incumbent big companies. Those big companies don’t see it as a threat — until years later, they are closing their doors. Every time a disruptive technology takes over, very few of the established players make it through to the other side. This does not guarantee that Google will dominate or crush those companies, or that everything that looks silly eventually wins. But it is a well established pattern.
This vehicle does not look threatening — not to people on the street, and not to existing car companies and pundits who don’t get it. Oh, there are many people inside those car companies who do get it, but the companies are incapable of getting it in their bones. Even when their CEOs get it, they can’t steer the company 90 degrees — there are too many entrenched forces in any large company. The rare exception are founder-led companies (like Google and Facebook and formerly Apple and Microsoft) where if the founder gets it, he or she can force the company to get it.
Even large companies who read this blog post and understand it still won’t get it, not most of the time. I’ve talked to executives from big car companies. They have a century of being car companies, and knowing what the means. Google, Tesla and the coming upstarts don’t.
One reason I will eventually move away from my chosen name for the technology — robocar — along with the other popular names like “self-driving car” is that this future vehicle is not a car, not as we know it today. It is no more a “driverless car” than a modern automobile is a horseless carriage. 100 years ago, the only way they could think of the car was to notice that there was no horse. Today, all many people notice about robocars is that no human is driving. This is the thing that comes after the car.
Some people expected the car to look more radical. Something like the Zoox or ATMBL by Mike and Maaike (who now work in a different part of Google.) Cars like those will come some day, but are not the way you learn. You start simple, and non threatening, and safe. And you start expensive — the Google prototype still has the very expensive Velodyne LIDAR on it, but trust me, very soon LIDAR is going to get a lot less expensive.
The low speed is an artifact of many things. You want to start safe, so you limit where you go and how fast. In addition, US law has a special exception from most regulations for electric vehicles that can’t go more than 25mph and stick to back roads. Some may think that’s not very useful (turns out they are wrong, it has a lot of useful applications) but it’s also a great way to start. Electric vehicles have another big advantage in this area. Because you can reverse electric motors, they can work as secondary brakes in the event of failure of the main brake system, and can even be secondary steering in case of failure of the steering system at certain speeds. (Google has also said that they have two steering motors in order to handle the risk of failure of one steering motor.) Electric vehicles are not long-range enough to work as taxis in a large area, but they can handle smaller areas just fine.
If you work in the auto industry, and you looked at this car and saw a clown car, that’s a sign you should be afraid.
Submitted by brad on Wed, 2014-05-28 00:40.
In what is the biggest announcement since Google first revealed their car project, it has announced that they are building their own car, a small low-speed urban vehicle for two with no steering wheel, throttle or brakes. It will act as a true robocar, delivering itself and taking people where they want to go with a simple interface. The car is currently limited to 25mph, and has special pedestrian protection features to make it even safer. (I should note that as a consultant to that team, I helped push the project in this direction.)
This is very different from all the offerings being discussed by the various car companies, and is most similar to the Navia which went on sale earlier this year. The Navia is meant as a shuttle, and up to 12 people stand up in it while it moves on private campus roads. It only goes 20 km/h rather than the 40 km/h of Google’s new car. Google plans to operate their car on public roads, and will have non-employees in test prototype vehicles “very soon.”
This is a watershed moment and an expression of the idea that the robocar is not a car but the thing that comes after the car, as the car came after the horse. Google’s car is disruptive, it seems small and silly looking and limited if you look at it from the perspective of existing car makers. That’s because that’s how the future often looks.
I have a lot to say about what this car means, but at the same time, very little because I have been saying it since 2007. One notable feature (which I was among those pushing for inside) is a soft cushion bumper and windshield. Clearly the goal is always to have the car never hit anybody, but it can still happen because systems aren’t perfect and sometimes people appear in front of cars quickly making it physically impossible to stop. In this situation, cars should work to protect pedestrians and cyclists. Volvo and Autoliv have an airbag that inflates on the windshield bars, which are the thing that most often kills a cyclist. Of the 1.2 million who are killed in car accidents each year, close to 500,000 are pedestrians, mostly in the lower income nations. These are first steps in protecting them as well as the occupants of the car.
The car has 2 seats (side-by-side) and very few controls. It is a prototype, being made at first in small quantities for testing.
More details, and other videos, including a one of Chris Urmson giving more details, can be found at the new Google Plus page for the car. Also of interest is this interview with Chris.
I’m in Milan right now about to talk to Google’s customers about the car — somewhat ironic — after 4 weeks on the road all over Europe. 2 more weeks to go! I will be in Copenhagen, Amsterdam, London and NYC in the coming weeks, after having been in NYC, Berlin, Krakow, Toronto, Amsterdam, Copenhagen, Oslo, the fjords and Milan. In New York, come see me at Singularity U’s Exponential Finance conference June 10-11.
Submitted by brad on Mon, 2014-04-28 12:44.
News from Google’s project is rare, but today on the Google blog they described new achievements in urban driving and reported a number of 700,000 miles. The car has been undergoing extensive testing in urban situations, and Google let an Atlantic reporter get a demo of the urban driving which is worth a read.
You will want to check out the new video demo of urban operations:
While Google speakers have been saying for a while that their goal is a full-auto car that does more than the highway, this release shows the dedication already underway towards that goal. It is the correct goal, because this is the path to a vehicle that can operate vacant, and deliver, store and refuel itself.
Much of the early history of development has been on the highway. Most car company projects have a focus on the highway or traffic jam situations. Google’s cars were, in years past, primarily seen on the highways. In spite of the speed, highway driving is actually a much easier task. The traffic is predictable, and the oncoming traffic is physically separated. There are no cyclists, no pedestrians, no traffic lights, no stop signs. The scariest things are on-ramps and construction zones. At low speed the highway could even be considered a largely solved problem by now.
Highway driving accounts for just over half of our miles, but of course not our hours. A full-auto car on the highway delivers two primary values: Fewer accidents (when delivered) and giving productive time back to the highway commuter and long distance traveller. This time is of no small value, of course. But the big values to society as a whole come in the city, and so this is the right target. The “super-cruise” products which require supervision do not give back this time, and it is debatable if they give the safety. Their prime value is a more relaxing driving experience.
Google continues to lead its competitors by a large margin. (Disclaimer: They have been a consulting client of mine.) While Mercedes — which is probably the most advanced of the car companies — has done an urban driving test run, it is not even at the level that Google was doing in 2010. It is time for the car makers to get very afraid. Major disruption is coming to their industry. The past history of high-tech disruptions shows that very few of the incumbent leaders make it through to the other side. If I were one of the car makers who doesn’t even have a serious project on this, I would be very afraid right now.
Submitted by brad on Mon, 2014-04-21 13:24.
Many states and jurisdictions are rushing to write laws and regulations governing the testing and deployment of robocars. California is working on its new regulations right now. The first focus is on testing, which makes sense.
Unfortunately the California proposed regulations and many similar regulations contain a serious flaw:
The autonomous vehicle test driver is either in immediate physical control of the vehicle or is monitoring the vehicle’s operations and capable of taking over immediate physical control.
This is quite reasonable for testing vehicles based on modern cars, which all have steering wheels and brakes with physical connections to the steering and braking systems. But it presents a problem for testing delivery robots or deliverbots.
Delivery robots are world-changing. While they won’t and can’t carry people, they will change retailing, logistics, the supply chain, and even going to the airport in huge ways. By offering very quick delivery of every type of physical goods — less than 30 minutes — at a very low price (a few pennies a mile) and on the schedule of the recipient, they will disrupt the supply chain of everything. Others, including Amazon, are working on doing this by flying drone, but for delivery of heavier items and efficient delivery, the ground is the way to go.
While making fully unmanned vehicles is more challenging than ones supervised by their passenger, the delivery robot is a much easier problem than the self-delivering taxi for many reasons:
- It can’t kill its cargo, and thus needs no crumple zones, airbags or other passive internal safety.
- It still must not hurt people on the street, but its cargo is not impatient, and it can go more slowly to stay safer. It can also pull to the side frequently to let people pass if needed.
- It doesn’t have to travel the quickest route, and so it can limit itself to low-speed streets it knows are safer.
- It needs no windshield or wheel, and can be small, light and very inexpensive.
A typical deliverbot might look like little more than a suitcase sized box on 3 or 4 wheels. It would have sensors, of course, but little more inside than batteries and a small electric motor. It probably will be covered in padding or pre-inflated airbags, to assure it does the least damage possible if it does hit somebody or something. At a weight of under 100lbs, with a speed of only 25 km/h and balloon padding all around, it probably couldn’t kill you even if it hit you head on (though that would still hurt quite a bit.)
The point is that this is an easier problem, and so we might see development of it before we see full-on taxis for people.
But the regulations do not allow it to be tested. The smaller ones could not fit a human, and even if you could get a small human inside, they would not have the passive safety systems in place for that person — something you want even more in a test vehicle. They would need to add physical steering and braking systems which would not be present in the full drive-by-wire deployment vehicle.
Testing on real roads is vital for self-driving systems. Test tracks will only show you a tiny fraction of the problem.
One way to test the deliverbot would be to follow it in a chase car. The chase car would observe all operations, and have a redundant, reliable radio link to allow a person in the chase car to take direct control of any steering or brakes, bypassing the autonomous drive system. This would still be drive-by-wire(less) though, not physical control.
These regulations also affect testing of full drive-by-wire vehicles. Many hybrid and electric cars today are mostly drive-by-wire in ordinary operations, and the new Infiniti Q50 features the first steer-by-wire. However the Q50 has a clutch which, in the event of system failure, reconnects the steering column and the wheels physically, and the hybrids, even though they do DBW regenerative braking for the first part of the brake pedal, if you press all the way down you get a physical hydraulic connection to the brakes. A full DBW car, one without any steering wheel like the Induct Navia, can’t be tested on regular roads under these regulations. You could put a DBW steering wheel in the Navia for testing but it would not be physical.
Many interesting new designs must be DBW. Things like independent control of the wheels (as on the Nissan Pivo) and steering through differential electric motor torque can’t be done through physical control. We don’t want to ban testing of these vehicles.
Yes, teams can test regular cars and then move their systems down to the deliverbots. This bars the deliverbots from coming first, even though they are easier, and allows only the developers of passenger vehicles to get in the game.
So let’s modify these regulations to either exempt vehicles which can’t safely carry a person, or which are fully drive-by-wire, and just demand a highly reliable DBW system the safety driver can use.
Submitted by brad on Tue, 2014-04-08 22:35.
I read a lot of feeds, and there are now scores of stories about robocars every week. Almost every day a new publication gives a summary of things. Here, I want to focus on things that are truly new, rather than being comprehensive.
Mahindra “Rise” Prize
The large Indian company Mahindra has announced a $700,000 Rise prize for robocar development for India’s rather special driving challenges. Prizes have been a tremendous boost to robocar development and DARPA’s contests changed the landscape entirely. Yet after the urban challenge, DARPA declared their work was done and stopped, and in spite of various efforts to build a different prize at the X-Prize foundation, the right prize has never been clear. China has annual prizes and has done so for several years, but they get little coverage outside of China.
An Indian prize has merit because driving in India is very much different, and vastly more chaotic than most of the west. As such, western and east Asian companies are unlikely to spend a lot of effort trying to solve the special Indian problems first. It makes sense to spur Indian development, and of course there is no shortage of technical skill in India.
Many people imagine that India’s roads are so chaotic that a computer could never drive on them. There is great chaos, but it’s important to note that it’s slow chaos, not fast chaos. Being slow makes it much easier to be safe. Safety is the hard part of the problem. Figuring out just what is happening, playing subtle games of chicken — these are not trivial, but they can be solved, if the law allows it.
I say if the law allows it because Indians often pay little heed to the traffic law. A vehicle programmed to strictly obey the law will probably fail there without major changes. But the law might be rewritten to allow a robot to drive the way humans drive there, and be on an open footing. The main challenge is games of chicken. In the end, a robot will yield in a game of chicken and humans will know that and exploit it. If this makes it impossible for the robot to advance, it might be programmed to “yield without injury” in a game of chicken. This would mean randomly claiming territory from time to time, and if somebody else refuses to yield, letting them hit you, gently. The robot would use its knowledge of physics to keep the impact low enough speed to cause minor fender damage but not harm people. If at fault, the maker of the robot would have to pay, but this price in damage to property may be worthwhile if it makes the technology workable.
The reason it would make things workable is that once drivers understood that, at random, the robot will not yield (especially if it has the right-of-way) and you’re going to hit it. Yes, they might pay for the damage (if you had the right of way) but frankly that’s a big pain for most people to deal with. People might attempt insurance fraud and deliberately be hit, but they will be recorded in 3D, so they had better be sure they do it right, and don’t do it more than once.
Of course, the cars will have to yield to pedestrians, cylists and in India, cows. But so does everybody else. And if you just jump in front of a car to make it hit the brakes, it will be recording video of you, so smile.
New Vislab Car
I’ve written before about Vislab at the University of Parma. Vislab are champions of using computer vision to solve the driving problem, though their current vehicles also make use of LIDAR, and in fact they generally agree with the trade-offs I describe in my article contrasting LIDAR and cameras.
They have a new vehicle called DEEVA which features 20 cameras and 4 lasers. Like so many “not Google” projects, they have made a focus on embedding the sensors to make them not stand out from the vehicle. This continues to surprise me, because I have very high confidence that the first customers of robocars will be very keen that they not look like ordinary cars. They will want the car to stand out and tell everybody, “Hey, look, I have a robocar!” The shape of the Prius helped its sales, as well as its drag coefficient.
This is not to say there aren’t people who, when asked, will say they don’t want the car to look too strange, or who say, looking at various sensor-adorned cars, that these are clearly just lab efforts and not something coming soon to roads near you. But the real answer is neither ugly sensors nor hidden sensors, but distinctive sensors with a design flair.
More interesting is what they can do with all those cameras, and what performance levels they can reach.
I will also note that car uses QNX as its OS. QNX was created by friend I went to school with in Waterloo, and they’re now a unit of RIM/Blackberry (also created by classmates of mine.) Go UW!
Submitted by brad on Sat, 2014-03-15 15:09.
One sign of how interest is building is the large reaction to some recent concept prototypes for robocars, two of which were shown in physical form at the Geneva auto show.
The most attention came to the Swiss auto research company Ringspeed’s XchangE concept which they based on a Tesla. They including a steering wheel which could move from side to side (and more to the point, go to the middle, where it could be out of the way of the two front seats,) along with seats that could recline to sleeping positions or for watching a big-screen TV, and which could reverse for face-to-face seating.
Also attracting attention was the Link and Go, an electric shuttle. In this article it is shown on the floor with the face to face configuration.
This followed on buzz late last year over the announcement of Zoox and their Boz concept, which features a car that has no steering wheel, and is symmetrical front to back (so of course seating is face to face.) The Zoox model takes this down to the low level, with 4 independent wheel motors. I’ve met a few times with Zoox’s leader, Tim Kentley-Klay of Melbourne, and the graphics skills of he and his team, along with some dynamic vision, also generated great buzz.
All this buzz came even though none of these companies had anything to say about the self-driving technology itself, which remains 99% of the problem. And there have been a number of designers who have put out graphic concepts like these for many years, and many writers (your unhumble blogger included) who have written about them for years.
The Zoox design is fairly radical — a vehicle with no windshield and no steering wheel — it can never be manually driven and a full robocar. Depending on future technologies like cheap carbon fibre and cost-effective 3-D printing for medium volumes, it’s a more expensive vehicle that you could make, but there may be a certain logic to that. Tesla has shown us that there are many people who will happily pay a lot more to get a car that is unlike any other, and clearly the best. They will pay more than can be rationally justified.
Speaking of Tesla, a lot of the excitement around the Rinspeed concept was that it was based on a Tesla. That appears to have been a wise choice for Rinspeed as people got more excited about it than any other concept I’ve seen. The image of people reclining, watching a movie, brought home an image that has been said many times in print but not shown physically to the world in the same way.
It’s easy for me (and perhaps for many readers of this blog) to feel that these concepts are so obvious that everybody just gets them, but it’s clearly not true. This revolution is going to take many people by surprise.
Submitted by brad on Thu, 2014-03-13 15:59.
Tuesday, the California DMV held a workshop on how they will write regulations for the operation of robocars in California. They already have done meetings on testing, but the real meat of things will be in the operation. It was in Sacramento, so I decided to just watch the video feed. (Sadly, remote participants got almost no opportunity to provide feedback to the workshop, so it looks like it’s 5 hours of driving if you want to really be heard, at least in this context.)
The event was led by Brian Soublet, assistant chief counsel, and next to him was Bernard Soriano, the deputy director. I think Mr. Soublet did a very good job of understanding many of the issues and leading the discussion. I am also impressed at the efforts Mr. Soriano has made to engage the online community to participate. Because Sacramento is a trek for most interested parties, it means the room will be dominated by those paid to go, and online engagement is a good way to broaden the input received.
As I wrote in my article on advice to governments I believe the best course is to have a light hand today while the technology is still in flux. While it isn’t easy to write regulations, it’s harder to undo them. There are many problems to be solved, but we really should see first whether the engineers who are working day-in and day-out to solve them can do that job before asking policymakers to force a solution. It’s not the role of the government to forbid theoretical risks in advance, but rather to correct demonstrated harms and demonstrated unacceptable risks once it’s clear they can’t be solved on the ground.
With that in mind, here’s some commentary on matters that came up during the session.
How do the police pull over a car?
Well, the law already requires that vehicles pull over when told to by police, as well as pull to the right when any emergency vehicle is passing. With no further action, all car developers will work out ways to notice this — microphones which know the sound of the sirens, cameras which can see the flashing lights.
Developers might ask for a way to make this problem easier. Perhaps a special sound the police car could make (by holding a smartphone up to their PA microphone for example.) Perhaps the police just reading the licence plate to dispatch and dispatch using an interface provided by the car vendor. Perhaps a radio protocol that can be loaded into an officer’s phone. Or something else — this is not yet the time to solve it.
It should be noted that this should be an extremely unlikely event. The officer is not going to pull over the car to have a chat. Rather, they would only want the car to stop because it is driving in an unsafe manner and putting people at risk. This is not impossible, but teams will work so hard on testing their cars that the probability that a police officer would be the first to discover a bug which makes the car drive illegally is very, very low. In fact, not to diminish the police or represent the developers as perfect, but the odds are much greater that the officer is in error. Still, the ability should be there. read more »
Submitted by brad on Tue, 2014-03-11 14:40.
Last year, the NHTSA released a document defining “levels” from 0 to 4 for self-driving technology. People are eager for taxonomy that lets them talk about the technology, so it’s no surprise that use of the levels has caught on.
The problem is that they are misleading and probably won’t match the actual progress of technology. That would be tolerable if it weren’t for the fact that NHTSA itself made recommendations to states about how the levels should be treated in law, and states and others are already using the vocabulary in discussing regulations. Most disturbingly, NHTSA recommendations suggested states hold off on “level 4” in writing regulations for robocars — effectively banning them until the long process of un-banning them can be done. There is a great danger the levels will turn into an official roadmap.
Because of this, it’s worth understanding how the levels are already incorrect in the light of current and soon-to-be-released technology, and how they’re likely to be a bad roadmap for the future.
Read A Critique of the NHTSA and SAE “Levels” for robocars.
Submitted by brad on Thu, 2014-03-06 14:41.
I often see the suggestion that as Robocars get better, eventually humans will be forbidden from driving, or strongly discouraged through taxes or high insurance charges. Many people think that might happen fairly soon.
It’s easy to see why, as human drivers kill 1.2 million people around the world every year, and injure many millions more. If we get a technology that does much better, would we not want to forbid the crazy risk of driving? It is one of the most dangerous things we commonly do, perhaps only second to smoking.
Even if this is going to happen, it won’t happen soon. While my own personal prediction is that robocars will gain market share very quickly — more like the iPhone than like traditional automotive technologies — there will still be lots of old-style cars around for many decades to come, and lots of old-style people. History shows we’re very reluctant to forbid old technologies. Instead we grandfather in the old technologies. You can still drive the cars of long ago, if you have one, even though they are horribly unsafe death traps by today’s standards, and gross polluters as well. Society is comfortable that as market forces cause the numbers of old vehicles to dwindle, this is sufficient to attain the social goals.
There are occasional exceptions, though usually only if they are easy to do. You do have to install seatbelts in a classic car that doesn’t have them, as well as turn signals and the other trappings of being street legal.
While I often talk about the horrible death toll, and how bad human drivers are, the reality is that this is an aggregation over a lot of people. A very large number of people will never have an accident in their lives, let alone one with major injuries or death. That’s a good thing! The average person probably drives around 600,000 miles in a lifetime in the USA. There is an accident for every 250,000 miles, but these are not evenly distributed. Some people have 4 or 5 accidents, and many have none.
As such, forbidding driving would be a presumption of guilt where most are innocent, and tough call from a political standpoint.
That doesn’t mean other factors won’t strongly discourage driving. You’ll still need a licence after all, and that licence might get harder and harder to get. The USA is one of the most lax places in the world. Many other countries have much more stringent driving tests. The ready availability of robotaxis will mean that many people just never go through the hassle of getting a licence, seeing no great need. Old people, who currently fight efforts to take away their licences, will not have the need to fight so hard.
Insurance goes down, not up
You will also need insurance. Today we pay about 6 cents/mile on average for insurance. Those riding in safe robocars might find that cost down to a penny/mile, which would be a huge win. But the cost for those who insist to drive is not going to go up because of robocars, unless you believe the highly unlikely proposition that the dwindling number of humans will cause more or deadlier accidents per person in the future. People tolerate that 6 cent/mile cost today, and they’ll tolerate it in the future if they want to. The cost will probably even drop a bit, because human driven cars will have robocar technologies and better passive safety (crumple zones) that make them much safer, even with a human at the wheel. Indeed, we may see many cars which are human driven but “very hard” to crash by mistake.
The relative cost of insurance will be higher, which may dissuade some folks. If you are told, “This trip will cost $6 if you ride, and $8 if you insist on driving” you might decide not to drive because 33% more cost seems ridiculous — even though today you are paying more for that cost on an absolute scale.
Highly congested cities will take steps against car ownership, and possibly driving. In Singapore, for example, you can’t have a car unless you buy a very expensive certificate at auction — these certificates cost as much as $100,000 for ten years. You have to really want a private car in Singapore, but still many people do.
Governments won’t have a great incentive to forbid driving but they might see it as a way to reduce congestion. Once robocars are packing themselves more tightly on the roads, they will want to give the human driven cars a wider berth, because they are less predictable. As such, the human driver takes up more road space. They also do more irrational things (like slow down to look at an accident.) One can imagine charges placed on human drivers for the extra road congestion they cause, and that might take people out of the driver’s seat.
The all-robocar lane tricks
There are certain functions which only work or only work well if all cars are robocars. They will be attractive, to be sure, but will they surpass the pressure from the human lobby?
- It’s possible to build dynamic intersections without traffic lights or bridges if all cars are trusted robocars.
- It’s possible to build low-use roads that are just two strips of concrete (like rails) if only robocars go on them, which is much cheaper.
- It’s possible to safety redirect individual lanes on roads, without need for barriers, if all cars in the boundary lanes are robocars. Humans can still drive in the non-boundary lanes pretty safely.
- We can probably cut congestion a lot in the all-robocar world, but we still cut it plenty as penetration increases over time.
These are nice, but really only a few really good things depend on the all-robocar world. Which is a good thing, because we would never get the cars if the benefits required universal adoption.
But don’t have an accident…
All of this is for ordinary drivers who are free of accidents and tickets. This might all change if you have an accident or get lots of tickets. Just as you can lose you licence to a DUI, I can foresee a system where people lose their licence on their first accident, or certainly on their second. Or their first DUI or certain major tickets. In that world, people will actually drive with much more caution, having their licence at stake for any serious mistake. A teen who causes an accident may find they have to wait several years to re-try getting a licence. It’s also possible that governments would raise the driving age to 18 or 21 to get people past the reckless part of their lives, but that this would not be a burden in a robocar world, with teens who are not even really aware of what they are missing.
I’ve driven over 35 years and had no accidents. I’ve gotten 2 minor speeding tickets, back in the 80s — though I actually speed quite commonly, like everybody else. It seems unlikely there would be cause to forbid me to drive, even in a mostly robocar world. Should I wish it. I don’t actually wish it, not on city streets. I still will enjoy driving on certain roads I would consider “fun to drive” in the mountains or by the coast. It’s also fun to go to a track and go beyond even today’s street rules. I don’t see that going away.
Submitted by brad on Mon, 2014-03-03 12:07.
In my recent travels, I have often been asked what various government entities can and should do related to the regulation of robocars. Some of them want to consider how to protect public safety. Most of them, however, want to know what they can do to prepare their region for the arrival of these cars, and ideally to become one of the leading centres in the development of the vehicles. The car industry is about to be disrupted, and most of the old players may not make it through to the new world. The ground transportation industry is so huge (I estimate around $7 trillion globally) that many regions depend on it as a large component of their economy. For some places it’s vital.
But there are many more questions than that, so I’ve prepared an essay covering a wide variety of ways in which policymakers and robocars will interact.
Read: Governments, The Law and Robocars
Submitted by brad on Mon, 2014-02-17 10:06.
It was revealed earlier this month that NHTSA wishes to mandate vehicle to vehicle radios in all cars. I have written extensively on the issues around this and regular readers will know I am a skeptic of this plan. This is not to say that I don’t think that V2V would not be useful for robocars and regular cars. Rather, I believe that its benefits are marginal when it comes to the real problems, and for the amount of money that must be spent, there are better ways to spend it. In addition, I think that similar technology can and will evolve organically, without a government mandate, or with a very minimal one. Indeed, I think that technology produced without a mandate or pre-set standards will actually be superior, cheaper and be deployed far more quickly than the proposed approach.
The new radio protocol, known as DSRC, is a point-to-point wifi style radio protocol for cars and roadside equipment. There are many applications. Some are “V2V” which means cars report what they are doing to other cars. This includes reporting one’s position tracklog and speed, as well as events like hitting the brakes or flashing a turn signal. Cars can use this to track where other cars are, and warn of potential collisions, even with cars you can’t see directly. Infrastructure can use it to measure traffic.
The second class of applications are “V2I” which means a car talks to the road. This can be used to know traffic light states and timings, get warnings of construction zones and hazards, implement tolling and congestion charging, and measure traffic.
This will be accomplished by installing a V2V module in every new car which includes the radio, a connection to car information and GPS data. This needs to be tamper-proof, sealed equipment and must have digital certificates to prove to other cars it is authentic and generated only by authorized equipment.
Robocars will of course use it. Any extra data is good, and the cost of integrating this into a robocar is comparatively small. The questions revolve around its use in ordinary cars. Robocars, however, can never rely on it. They must be be fully safe enough based on just their sensors, since you can’t expect every car, child or deer to have a transponder, ever.
One issue of concern is the timeline for this technology, which will look something like this:
- If they’re lucky, NHTSA will get this mandate in 2015, and stop the FCC from reclaiming the currently allocated spectrum.
- Car designers will start designing the tech into new models, however they will not ship until the 2019 or 2020 model years.
- By 2022, the 2015 designed technology will be seriously obsolete, and new standards will be written, which will ship in 2027.
- New cars will come equipped with the technology. About 12 million new cars are sold per year.
- By 2030, about half of all cars have the technology, and so it works in 25% of accidents. 3/4 of those will have the obsolete 2015 technology or need a field-upgrade. The rest will have soon to be obsolete 2022 technology. Most cars also have forward collision warning by this point, so V2V is only providing extra information in a tiny fraction of the 25% of accidents.
- By 2040 almost all cars have the technology, though most will have older versions. Still, 5-10% of cars do not have the technology unless a mandate demands retrofit. Some cars have the equipment but it is broken.
Because of the quadratic network effect, in 2030 when half of cars have the technology, only 25% of car interactions will be make use of it, since both cars must have it. (The number is, to be fair, somewhat higher as new cars drive more than old cars.) read more »
Submitted by brad on Tue, 2014-01-14 22:44.
I’m working on a new long article about advice to governments on how they should react to and encourage the development of robocars.
An interesting plan announced today has something I had not thought of: Michigan is funding the development of a fake downtown to act as a test track for robocar development. The 32 acre site will be at the University of Michigan, and is expected to open soon — in time for the September ITS World Congress.
Part of the problem with my advice to governments is that my main recommendation is to get out of the way. To not try too hard both to help and to regulate, because even those of us trying to build the vehicles don’t have a certain handle on the eventual form of the technology.
A test track is a great idea, though. Test tracks are hugely expensive to make, entirely outside of the means of small entrepreneurs. They immediately resolve most safety concerns for people just starting out — every team has had small runaway issues at the very start. Once past that, they can be shared, in fact having multiple vehicles running the track can be a bonus rather than a problem.
Big car companies all have their own test tracks, but these are mostly real tracks, not urban streets. Several companies have built pre-programmed robotic cars which drive in specific patterns to test ADAS systems and robocars. The DARPA Urban Challenge was run on an fake set of urban streets on an old military base, so this idea goes back to the dawn of the modern field. (Old military bases are popular for this — Mythbusters used a California one for their test of blind and drunk driving.)
This track will probably bring teams to Michigan, which is what they want. Detroit is in trouble, and it knows it. Robocars are going to upend the car industry. Incumbent players are going to fall, and new players are going to rise, and that could be very bad news for Detroit.
My home province of Ontario is facing the same problem, to a lesser degree. A lot of the Ontario economy is in cars as well, and so they’ve started a plan to introduce testing legislation. I don’t think this is the right plan — testing is already legal with a good supervising driver in most jurisdictions, though I have not yet examined the Ontario code. Ontario has one big advantage over Michigan, though, in that it is also a high-tech centre. Right now the car companies in Detroit are finding it very difficult to convince high-tech stars to come move to Detroit, in spite of being able to offer high pay and the fact that you can literally get a mansion for the price of the downpayment on a nice San Francisco condo. Toronto doesn’t have the same problem — in fact it’s one of the most desired places to live for Canadians, and for people from all over the world. Ontario’s combination of high-tech and big automotive might end up doing well.
At least in Ontario, everybody will be motivated to solve the snow problem sooner than the California companies are.
Submitted by brad on Fri, 2014-01-10 11:19.
With last week’s commercial release of the Navia, I thought I would release a new essay on the challenges of driving robocars at different speeds.
As the Navia shows, you can be safe if you’re slow. And several car company “traffic jam assist” products say the same thing. On the other end, we see demos taking place at highway speeds. But what about the middle range — decent speeds on urban streets?
Turns out that’s one of the harder problems, and so there is a “valley” in the chart which makes safe operation harder in that zone.
So read my more detailed essay on these challenges: The Valley of Danger for Robocars
Submitted by brad on Wed, 2014-01-08 17:44.
A significant milestone was announced this week. Induct has moved their “Navia” vehicle into commercial production, and is now taking orders, though at $250,000 you may not grab your wallet.
This is the first commercial robocar. Their page of videos will let you see it in operation in European pedestrian zones. It operates unmanned, can be summoned and picks up passengers. It is limited to a route and stops programmed into it.
The “catch” is that it stays safe by going only 20km/h, where it is much harder for it to harm things. It’s aimed at the campus shuttle market, rather than going on public roads, but it drives on ordinary pavement, not requiring special infrastructure, since it localizes using a prepared laser map of the route.
Now 20km/h (12mph) is not very fast, though suitable for a campus shuttle. This slow speed and limited territory may make some skeptical that this is an important development, but it is.
- This is a real product, ready to deploy with civilians, without its own dedicated track or modified infrastructure.
- The price point is actually quite justifiable to people who operate shuttles today, as a shuttle with human driver can cost this much in 1.5 years or less of operation.
- It smashes the concept of the NHTSA and SAE “Levels” which have unmanned operation as the ultimate level after a series of steps. The Navia is at the final level already, just over a constrained area and at low speed. If people imagined the levels were a roadmap of predicted progress, that was incorrect.
- Real deployment is teaching us important things. For example, Navia found that once in operation, teen-agers would deliberately throw themselves in front of the vehicle to test it. Pretty stupid, but a reminder of what can happen.
The low speed does make it much easier to make the vehicle safe. But now it become much easier to show that over time, the safe speed can rise as the technology gets better and better. (To a limit — see my article on the dangers at different speeds.)
The route limitation has two elements. The first is that they want to keep it only in safe locations, which makes sense for an early release. It also avoids legal issues. The second is simpler — they are using a map based approach, so they can only drive somewhere that has been mapped. Mapping means driving a scanner over the route and building a map of all the details, and then typically having humans confirm the map. This is the same way that the cars from Google and almost all other vendors do it when they are doing complex things that go beyond following lane markers on a highway. As such it is not that big a barrier. While building new infrastructure is hugely expensive, mapping it is much more modest in comparison, though non-trivial. Covering the whole world would take time, but it becomes possible to quickly add routes and destinations.
I single out the Navia because of its ability to drive without requiring any changes to the roads or extra infrastructure. Previous shuttle-style systems like the ULTra PRT at Heathow (which I rode a couple of months ago), the Masdar PRT and earlier Cybercar projects all required a dedicated guideway or fenced-off ground track to run. While the Navia is being kept to private property for safety and legal reasons, there is no technical reason it could not operate in public spaces, which moves it from PRT into Robocar territory.
The Navia is very much designed to be a shuttle. It is open-air and doesn’t really have seats, just padded bars to lean against. There is no steering wheel or other traditional control. This belies that common expectation of the first vehicles looking just like traditional cars.