Automated Vehicles Symposium Day 0: When do robocars become cheaper than standard cars?


I'm in the Detroit area for the annual TRB/AUVSI Automated Vehicle Symposium, which starts tomorrow. Today, those in Ann Arbor attended the opening of the new test track at the University of Michigan. Instead, I was at a small event with a lot of good folks in downtown Detroit, sponsored by SAFE which is looking to wean the USA off oil.

Much was discussed, but a particularly interesting idea was just how close we are getting to something I had put further in the future -- robocars that are cheaper than ordinary cars.

Most public discussion of robocars has depicted them as costing much more than regular cars. That's because the cars built to date have been standard cars modified by placing expensive computers and sensors on them. Many cars use the $75,000 Velodyne Lidar and the similarly priced Applanix IMU/GPS, and most forecasts and polls have imagined the first self-driving cars as essentially a Mercedes with $10,000 added to the price tag to make it self driving. After all, that's how things like Adaptive Cruise Control and the like are sold.

Google is showing us an interesting vision with their 3rd generation buggy-style car. That car has no steering wheel, brakes or gas pedal, and it is electric and small. It's a car aimed at "Mobility on Demand."

When people have asked me "how much extra will these cars cost," my usual answer has been that while the cars might cost more, they will be available for use by the mile, where they can cost less per mile than owning a car does today -- ie. that overall it will be cheaper. That's in part because of the savings from sharing, and having vehicles go more miles in their lifetime. More miles in the life of a car at the same cost means a lower cost per mile, even if the car costs a little more.

The sensors cost money, but that cost is already in serious decline. We're just a few years away from $250 Lidars and even cheaper radar. Cameras are already cheap, and there are super cheap IMUs and GPSs already getting near the quality we need. Computers of course get cheaper every year.

This means we are not too far when the cost of the sensors is less than the money saved by what you take out of the car. After all, having a steering wheel, gas and brakes costs money. Side mirrors cost money (ever had to replace them?) That fancy dashboard with all its displays and controls costs a lot of money, but almost everything it does in a robocar can be done by your tablet.

That said, you need a few extra things in your robocar. You need two steering motors and two braking systems. You need some more short range sensors and a cell phone radio. But there's even more you can save, especially with time.

Because mobility on demand means you can make cars that are never used for anything but short urban trips (the majority of trips, as it turns out) you can save a lot more money on those cars. These cars need not be large or fast. They don't need acceleration. They won't ever go on the highway so they don't need to be safe at 60mph. Electric drive, as we discussed earlier, is great for these cars, and electric cars have far fewer parts than gasoline ones. Today, their batteries are too expensive, but everything else in the car is cheaper, so if you solve the battery cost using the methods I outlined Saturday we're saving serious money. And small one or two person cars are inherently cheaper to boot.

Of course, you need to make highway cars, and long-range 4WD SUVs to take people skiing. But these only need be a fraction of the cars, and people who use a mix of cars will see a big saving.

For a long time, we've talked about some day also removing many of the expensive safety systems from cars. When the roads become filled with robocars, you can start talking about having so few accidents you don't need all the safety systems, or the 1/3 of vehicle weight that is attributable to passive safety. That day is still far away, though cars like the Edison2 Very-Light-Car have done amazing things even while meeting today's crash tests. Companies like Zoox and other startups have pushed visions of completely redesigned cars, some of them at lower cost for a while. But this seems like it might become true sooner rather than later.

Evacuation in a hurricane

One participant asked how, if we only had 1/9th as many cars (as some people forecast, I suspect it's closer to 1/4) we would evacuate sections of Florida or similar places when a hurricane is coming. I think the answer is a very positive one -- simply enforce car pooling / ride sharing in the evacuation. While there is not a lot I think policymakers should do at this time, some simple mandates could help a lot in this arena. While people would not be able to haul as much personal property, it is very likely there would be more than enough seats available in robocars to evacuate a large population quickly if you fill all the seats in cars going out. Further, those cars can go back in to get more people if need be.

Filling those seats would actually get everybody out faster, because there would be far less traffic congestion and the roads would carry far more people per hour. In fact, that's such a good idea it could even be implemented today. When there's an evacuation, require all to use an app to register when they are almost ready to leave. If you have spare seats, you could not leave (within reason) until you picked up neighbours and filled the seats. With super-carpooling, everybody would get out very fast on much less congested roads. Those crossing the checkpoint on the way out with empty seats would be photographed and ticketed unless the app allowed them to leave like that, or the app records that it tried to reach the server and failed, or other mitigating circumstances. (This is all hours before the storm, of course, before there is panic, when people will do whatever they can.) Some storms might be so bad the cars are at risk. In that case, if the road capacity is enough, people could move out all the cars too, to protect them. But in most cases, it's the people that are the priority.

More tomorrow as the conference gets underway.


While the *normal* number of cars in an evacuation area may not be enough, the number of (the larger long distance types) cars that can be mobilized from neighbouring areas (self piloted, no need to have a driver) should compensate easily.

For removal of goods, U-Haul (et al) will certainly be happy to send in self-driving trucks with trailers that you can fill up and then have them deliver your stuff to your inlaws in the next state, again without you having to anything other than fill them up.

Presumably (in the US) FEMA may develop mobilization plans to incorporate these capabilities.

I can imagine that during a disaster such as a hurricane, while the roads going away from areas that are expected to be affected will be packed, the roads going back in which would normally be empty will then wind up seeing use by automated vehicles going back in to the threatened area in order to be available to pick up more people and get them out.

This raises issues of the government passing laws that essentially conscript those vehicles from the private corporations that own fleets of them, as well as providing reimbursement for damage/destruction done to those vehicles that don't make it out in time.

Additionally there should be regulations put into place against fare gouging for the transportation of people who don't own their own cars/etc. Better yet, once there's an evacuation order the fares should simply be suspended, and the companies reimbursed for the lost revenue.

There's also going to need to be regulations or laws to prevent people from hoarding rides out of town, and filling seats with personal belongings rather than people.

Lots of socialist/anti-free market type regulations, in all this. Honestly I don't think these types of laws will actually get made in the U.S., at least until there are two or more big natural disasters where an inordinate number of poor people die because they didn't have their own vehicles, and the self-driving car companies both jacked up their rates and started gouging, as well as wouldn't send their cars back in to the evacuation area to pick up more people. (Yeah, the cars may be insured, but what business wouldn't rather not have their property be put in potential danger of damage or loss vs having to deal with insurance claims and potential constrained production? Avoiding a disaster entirely means no loss of business to wait for replacement assets even temporarily.)

Then the logistics or all this starts looking like FedEx package delivery, or Netflix physical media distribution & return. Who's going to write that logistics software, and get all the car companies to integrate it into their systems, when there's no direct profit in it? ("Well, if we just make it open source..." Hahahaha!)

In the robocar world, I anticipate there being lots of capacity. While most towns would not necessarily have enough seats to carry everybody at once, the ability to summon vehicles in an instant from other towns should cause a very large supply, and thus prices should not get all that high.

If there is a shortage, you could have a rule fining people who come out with empty seats during the declared emergency and shortage.

You would have different rules based on how imminent the disaster is, and for different ones. Fire you want more of your stuff, for example but it may be more urgent. All automated transit buses for miles around would be put into service. You could flood an area with vast numbers of cars in just a couple of hours with an emergency order. The price gouging would not be in the emergency area, but in the areas that lost their cars because the emergency order sent them all into the evacuation area.

Add new comment