Car design changes due to Robocars


Robocars will suggest a great number of possible changes in the way we design and market cars. I now encourage you to read:

Automobile design changes due to Robocars

The big green benefit of robocars comes in large part from the freedom they offer in redesigning the automobile, in particular the ability to specialize automobiles to specific tasks, because they can be so readily hired on demand. Or to specific fuels in certain areas, or for sleeping, and much more.


Brad. If I wanted a car like this I would take a L'Eggs Pantyhose container, enlarge it, and glue it to my bike.

I truly hope this is not your car. You are much sexier than this thing!

It is yours, isn't it (((hangs head in shame)))

I will be forced to find a different guru.

I don't have one, it's an electric tricycle. It's an example of a super efficient ultralightweight, which uses about 100 btus/mile, of 250 at the power plant, which is 10-20x as efficient as cars or transit systems.

It's just an example of the type of vehicle that might be designed when you can design a vehicle for a specific purpose (in this case efficient short urban trips for a single person.)

Most trips over 10 miles involve the use of highways.

Highways are a tiny fraction of the road miles in the United States.

The power and energy storage requirements for electric vehicles are dominated by the problems associated with highway driving.

Highways almost never have people walking around on them. People walking around on highways are already at very high risk of being killed, and this risk is widely understood.

Therefore, it would be reasonable to deliver electricity from metal strips flush with the surface of the highway to the cars on the highway. Cars with pickups underneath (similar to the overhead pantographs on electric trains) would not need batteries anywhere near as large as cars without, and would have unlimited range.

The same pickups would be used in garages where users would install charging devices into the floor.

An APS system ( could be used in a few areas where pedestrians are possible, but either cars need large amounts of power (San Francisco's hills?) or range boosting (El Camino Real, down the spine of the San Francisco Peninsula).

While such things are possible, I doubt they will happen because they require "big infrastructure" and anything involving that happens very slowly and is rarely highly innovative. It's one of the reasons I think robocars are the likely answer because while they require lots of impressive innovation they require no infrastructure. Once you have a robocar one person can buy it and use it to go anywhere that it's legal to take it. Magic highways only work where you raise the billions to build the magic highway for those with the special car.

Infrastructure changes do happen of course, but so slowly that anything that can be done by small but clever innovative teams wins.

According to Rough Guides, the US road system has over 5.7 million miles of roads. Doing anything to a system that size is, as you say, big infrastructure, and takes forever.

But look at this, especially page 2:

There are just 15,300 miles of urban interstate highway. 24% of travel in the US occurs on just 1.2% of the road system. Some very high fraction of longer trips use the interstate system along part of the length. This is a huge lever.

At $200/foot, it would cost just $16 billion to put electrical track into the whole system. And note that there is no reason to do it all at once. 250 miles ($264 million) would open a large market in the San Francisco Bay Area. Twice that would open a much larger market in Los Angeles.

Consider the size of the market. The Bay Area population is 7 million people. A the average rate of US car ownership, that's 5.7 million vehicles. Vehicles have a life of around 16 years, so that's 356,000 new vehicles a year required in the Bay Area. If 10% of the market purchased cars with electrical pickups, the infrastructure would be serving over 260,000 vehicles in 8 years, at an infrastructure cost of under $1000/vehicle. For comparison, the minimum avoided cost of a 16 kW-hr battery (minimum for all-electric government designation) is $3600 (lead-acid, 100% of cycle, totally impractical).

I think your vision of the future of transportation being a lot like the present (individual cars) is compelling, and I agree that increased robotics in the car is inevitable. I think the powered roadway idea solves some of the biggest transport problems that we have (pollution, oil imports, global warming), on a shorter timeline and with less acceptance hurdles than robot cars.

While I think it's possible to design it safely, people will actually be pretty scared of electrification of roadway, even roadway that pedestrians are not allowed on.

I also don't think you could do it for $200/foot. That's just a million per mile and highway work tendsd to be way north of that.

I actually blogged on this topic here:

I got started on the idea by a South Korean project that was running the cars by inductive power transfer and actually quite safe. There was actually an extensive discussion about it here:

I kind of ran out of enthusiasm for the idea after I calculated how much power all the electric cars would consume during afternoon rush hour. Assuming 10kw per car, I came up with 170GW. Since afternoon rush hour coincides with peak electrical power consumption, there is no way to supply that much power without building a lot of new power plants. I guess I should post parts 2 and 3 and close out the topic, but I discuss all the details over in the Energy from Thorium forum. The trouble was the power plants would have cost 10 times what the road would and that was using the cheapest Natural Gas peaking power.

From an environmental point of view it wasn't clear to me that this would actually be beneficial compared to running the cars off CNG in the first place.

I'm now interested in development work on the lithium-air battery. If we can develop a reasonably priced battery that store 1 KWh per KG, then it is a whole new ball game. That would use similar amounts of electricity, but using off peak power, wouldn't require new power plants immediately, although in the long term you would want to build nuclear power plants to reduce CO2 emissions.

Do you guess that in the future will be used only cars with electricity power? Or will be there a regulation for only electric cars?


Glad I found someone who can start to see the huge impact of robotic cars. You mention designing more closely for function, but we also need to start designing for robotic manufacturing, in total, and for robotic maintenance. This allows the vehicle to be maintained when it is waiting and getting fueled and whatever. The Army has it started if you look at changing out the power plants for tanks and their more modern heavy haulers.

Also, I don't think waiting for fully autonomous cars is necessary. The biggest issue is keeping the on the road in any weather, so using a system of magnetic strips embedded in the road, cheap to add, could provide micro and macro positional data. This lets the car use sensors for its perimeter protection. Gigabit Ethernet can let cars communicate to avoid each other and cooperate. Intersection Controllers can keep all cars in their area interlacing properly with minimum speed changes.

The impact on infrastructure costs will be huge. Currently on the busiest traffic only about 18% of a freeway is used. With robocars we can triple that and still allow emergency room. That means a 5 lane highway drops to 2. That means land can be returned to other uses and maintenance costs are cut by the same margin. The impact on bridge design, building, and maintenance are hugely curtailed. Sidewalks and greenspaces in cities get larger and fewer peds are whacked each year. Trees have more room to grow and grass can be line the roads because there is far more space because lanes are far narrower.

BTW, The Heathrow Transit system based on robotic/computer controller cars show the effects you mention.

Early experiments in robocars relied on things like special highways with magnetic strips. A bus experiment in Berkeley also did this, the driver handled the gas and brake but not the steering, and the big win was the bus always parked so perfectly at the curb you could roll wheelchairs right in over the tiny gap.

But that's old-think. Truth is, knowing the boundaries of the road is a minor part of the problem, and one we can solve other ways, much more cheaply than installing stuff on all our roads. I think it is essential that the cars be able to go anywhere, at least anywhere they have a map of, and we already have maps of just about everywhere (though we need to improve their reliability but that's actually not that hard if you get logs of where people drive.)

I'm not against the magnets -- the robot should use every tool it can get -- but I am against their being necessary, that I can't go down my own street until somebody buries the magnets. And then what if they fail?

However, I do agree about robotic maintenance being a good idea, though robotic recharge/refuel is a good start.

There may be some very good reasons to embed information into the road surface whether it be in the form of magnetic strips, RF or some other sort of passive information source. The major reason is that it offers a robust source of information. (Foreign hackers could break into and change information in an IT network, but trying to remove large numbers of devices stamped into a road surface would be more of a challenge). There is no reason to think that machines could not embed devices as quickly as they paint white lines. Devices that give information such as exact position, speed limit, acceleration rate and directions to the next device(s) would give strong backup to GPS and stored data.

Well, you can hack just about anything, including magnetic fields, which is why you don't want to depend on any one thing and really you want to depend on the thing humans depend on, namely understanding the world through sensors, be they vision or laser or radar or whatever, and all of them. You can throw in magnets, but if you say "No, your robocar can only drive on streets where we embedded the magnets" I think you seriously hurt the market for it, and you take away the self-delivering benefit which is one of the largest benefits, unless there are magnets guiding to where you are going.

Special infrastructure could work for a very short time to declare special roads for empty vehicles, but it's a short term solution.

There is a world of difference between a localized hack of passive markers and a large systems hack. You can hack the white line on the road if you are good with a paint brush, or an expert could theoretically hack into the GPS system.
I agree an intelligent car needs multiple sources of information to verify its course.
I did not say "our robocar can only drive on streets where we embedded the magnets". Just as humans are still able to drive when passive markers such as roadside reflectors or white lines are not available, I would expect the same from robocar. However just like humans, some driving adjustments are often necessary when there is less information about the road parameters.

I understand you can put in the magnets and not be required to depend on them. I am all in favour of the robot having every piece of data it can get its hands on to make decisions about where to go.

It's just that usually when people talk about putting in special infrastructure, it is with both the implication that the cars will only go where that is, and also the likely result that others will want such a restriction even if that wasn't intended originally.

Our liability system assigns a huge importance to the best possible safety. Bizarrely, this can mean that you can't have two level of safety, even if they are "really safe" and "even safer." Once you declare that the cars are safer on a street with magnetic markers, they will say, "If they are less safe on the other streets, let's hold off until we get the magnets in."

Google's announcement today shows just how good we're getting at making it work without changing the infrastructure, and I think that's the best path to deployment. And that, in turn, is the safest course, because once the vehicles are safer than people, they should be used everywhere, not just where they are even safer.

As most car journeys are made with just one person, placing a single seat in the centre of the car where it is safest and gives the passenger more room makes sense. (This is assuming many journeys in the future will be in driver-less taxis)
Access into the vehicle through a single door opening onto the sidewalk increases safety as it is the safer side to enter the vehicle. It also makes the vehicle cheaper to manufacture.
As visibility is not as important if you don't have to drive, the seat could be within a protective shell wrapping around the back and partially around the sides and extending to the roof. Some of the accidents racing car drivers walk away from show what is possible with vehicle safety.
As heating/cooling makes bigger demands on energy as cars get smaller (especially with all electric vehicles)it becomes more important to insulate the cabin. Possibly only the front and forward side windows may be necessary, the back window just increases heat loss and could be deleted.
As many people would be watching a screen of some sort instead of driving, heavily tinting and double glazing the remaining windows to darken the interior would help. As a bonus a vehicle like this would be very quiet inside.
The vehicle could also be more streamline as the boxy shape needed for headroom for side by side seating is not needed.
It should be possible to give the passenger much more room yet reduce the volume of the cabin by loosing space that is not used.
A show like "Top Gear" would no doubt bag a car like this, but in the long term we are going to be forced to make some pretty big changes.

Top Gear will hate just about everything to do with robocars, I suspect.

Some people won't mind a reduced visibility car, and others will hate it both for aesthetic reasons and motion sickness reasons. I think they will just use the much brighter screens we can build today rather than dimming the windows, unless it's easy to dim them.

I could see "convertible" designs where there is a top which quickly comes off. You send your car to a depot where they have various tops which fit the vehicle, and you can go with no top, a solid top, a transparent top, tinted top etc. The "top" might include the windows that aren't part of the door.

Getting to and from medical appointments is a major chore for millions of seniors. It occurs at the precise time in their lives when their driving abilities are failing. The rapid adoption of a limited automated medical taxi system will be a huge boon to both the seniors and other drivers. I can visualize healthcare systems having a fleet of vehicles to pick up patients just in time for their appointments. A slightly larger vehicle could have space for a medical practitioner to take blood pressure readings and other vital signs on the way to the clinic and to dispense advice and care requirements on the return trip. For that matter, a robo-medical-taxi could automate the blood pressure, weight, and other vital signs on the way to the appointment.

Individuals don't need to own cars:
Go to a subscription model, Subscribe to 10k miles per year, call a car when needed.
Multiple Benefits to this model:
1. eliminate double trips to go somewhere to pick some one up. Send a car to the school to pick up your kid and the car drives your kid back home.
2. A North American Suburban City of 10 MM residents has approximately 5 million cars, these cars sit around for 90% of the day. So on *average*, just 500k cars can satisfy the transportation needs of the 10 MM residents.
3. Eliminate the need for parking everywhere. Businesses, Schools and Shopping complexes don't need to provide for parking or reduce the need for parking. Homes don't need garages either.
4. The subscriber can choose the type of car as needed, Single passenger, Multi passenger, or cargo capacity.

Some will own. Some will own and hire out when not in use. Some will not own and always summon when needed. We'll see a drop in the number of cars, though probably not to 1/10th as you suggest. Perhaps to 1/4 though, depending on the blend above. However, total miles will not drop, and in fact will probably increase a bit.

Y'all are focusing on the wrong area here. Private cars will *not* be the primary application for driverless vehicles. Trucking will be. There's an ENORMOUS opportunity for cost savings if fleets can put robots into trucks and drive them 24 hours across the country.

I can also imagine couriers being impacted. You need to get something across town?
Just call a robo-taxi, put your goods in it and give it the destination.
The same could work for local retailers and e-tailers. You could track your delivery via GPS and know exactly when your order will arrive.

I really like your idea of chasing temporarily-parked robocars out of my driveway. Hyah! Git! Mush!

To DonAZ's point: there are something like 4M professional drivers in this country (3.2M truck, the rest taxi and limo), essentially all of whom are going to be out of work. Another line of work down the drain for those without college degrees.

By the way I still have the "I went to Mexico for a corona" button you sold me there in 1991.

Wow, that's deep history:

I don't think the driver count is quite that high, but I see the census count is up there. I think long haul trucking will be one of the last forms to be converted to automatic actually because the safety records of those drivers are good and the vehicles are so heavy.

I imagine that driver wages is a significant fraction of the price of trucking, making it desirable to drive fast (modulo legality and safety issues). Once that's gone, presumably fuel will be the next limiting factor, especially as it gets more expensive. It's more fuel-efficient to go slower. Get ready for a lot of crawling robotrucks, and pressure from trucking companies to lower the minimum speed on highways overnight.

Generally vehicles do have a most efficient speed. It's not super slow because there is both rolling resistance and drag to consider. There is already a move afoot to start converting trucks to compressed natural gas, which is much, much cheaper than gasoline, modestly cleaner, and domestic, and going through a boom. So I am not sure peak oil will affect the trucks that much. People just want their cargo there faster. Those trucks get about 5-6 mpg (of diesel) which means at 70mph they are burning fuel at about $50/hour at $4/gallon. The driver is probably getting $15 to $20 per hour.

What a great post. The density of new ideas is very high. Hope Google (and Ford, and GM, etc.) have read it. I wonder if Mr. Templeton has had any feedback from the major auto manufacturers on these ideas?

This seems to be a classic "Innovator's Dilemma" phenomenon, where the established car companies are minimally involved in the next phase of transportation- i.e., self-driving automobiles. Mr. Templeton lays out the many radically new form factors involved with the new cars. What strikes me is that these new cars will depend on legacy automotive technologies to a very small degree, thus setting the stage for large numbers of new entrants.

People are starting to realize that robo-cars remove range anxiety from electric vehicles. Energy efficiency of one thousand mpg equivalent is possible. See my paper from IEEEE Technology and Society:

Multi-passenger cars could coordinate pickups on the fly much as a radio dispatched airport shuttle does now but more like driverless Uber for commuters. Regular, semi-regular and ad hoc use would be handled gracefully.

These cars could drive closer in line reducing drag and road are used per car. This saves energy and reduces traffic. ON the highway, they could hook up and travel in a mutually beneficial draft group like NASCAR. The lead car may use more energy but could receive some of the overall savings as a stored on board energy credit.

All cars waiting at a stop light could accelerate simultaneously increasing the number of cars that make the light.

Add new comment