Robocars and electrification

One of my first rules of robocars is "you don't change the infrastructure." Changing infrastructure is very hard, very expensive, requires buy-in from all sorts of parties who are slow to make decisions, and even if you do change it, you then have a functionality that only works in the places you have managed to change it. New infrastructure takes many decades -- even centuries, to become truly ubiquitous.

That's why robocar enthusiasts have been skeptical of things like ITS plans for roadside to vehicle and vehicle to vehicle communications, plans for dedicated highway lanes with special markers, and for PRT which needs newly built guideways. You have to work with what you have.

There are some ways to bend this rule. Some infrastructure changes are not too hard -- they might just require something as simple and cheap as repainting. Some new infrastructures might be optional -- they make things better in the places you put them, but they are not necessary to operations. Some might focus on specific problem areas -- like special infrastructure in heavy pedestrian areas or parking lots, enabling or improving optional forms of operation in those areas.

Another possiblility is to have robocars enable a form of new infrastucture, turning it upside down. The infrastructure might need the robocars rather than the other way around. I wrote about that sort of plan when discussing a solar panel on a robocar.

A recent proposal from Siemens calls for having overhead electric wires for trucks. Trolley buses and trams use overhead electric wires, and there are hybrid trolley buses (like the Boston T line) which can run either on the wires or on an internal diesel. These trucks are of that type. The main plan for this is to put overhead wires in things like shipping ports, where trucks are running around all the time, and they would benefit greatly from this.

I've seen many proposals for electrication of the roads. Overhead wires are problematic because they need to be high enough to go over the trucks and other high vehicles, but that makes them harder to reach by low vehicles. You need two wires and must get good contact. They are also damn ugly. This has lead to proposals for inductive power supplies buried in the road. This is very expensive as it requires tearing up the road. There are also inductive losses, and while you don't need to make contact, precise driving is important for efficiency. In these schemes, battery-electric cars would be able to avoid using their batteries (and in fact charge them) while on the highway, vastly increasing their range and utility.

Robocars offer highly precise driving. This would make it easier to line up on overhead wires or inductive coils in the road. It even would make it possible to connect with rails in the roadbed, though right now people don't want to consider having a high voltage rail on the ground, even on a highway.

It was proposed to me (I'm trying to remember by who -- my apologies) that one new option would be a rail on the side of the highway. This lane would be right up against the guardrail, and normally would be the shoulder. In the guardrail would be power rails, and a connector would come from the left side of the vehicle. Only a robot would be able to drive so precisely as to do this safely. Even with a long pole and more distance I am not sure people would enjoy trying to drive like this. A grounding rail in the roadbed might also be an option -- though again tearing up the roadbed is very expensive to do and maintain.

There is still the problem of having a live rail or wire at reachable height. The system might be built with an enclosed master cable and then segments of live wire which are only live when a vehicle is passing by them. Obviously a person doesn't want to be there when a car is zooming through. This requires roboust switching eqiupment for the thousands of watts one wishes to transfer. You also have to face the potential that a car from the regular lanes could crash into the rail and wires, and while that's never going to be safe you don't want to make it worse. You also need switching if you are going to have accounting, so only those who pay for it get power. (Alternately it could be sold by a subscription so you don't account for the usage and you identify cars that don't have a subscriber tag who are sucking juice and fine them.)

There is also the problem that this removes the shoulder which provides safety to other cars and provides a breakdown lane. If a vehicle does have to stop in this lane for emergency reasons, sensors in the rail could make sure that all robocars would know and leave the lane with plenty of margin. They would all have batteries or engines and be able to operate off the power -- indeed the power lines need not be continuous, you don't have to build them in sections of the road where it's difficult. If other cars are allowed to enter the lane, it must not be dangerous other than physically for them to brush the wires.

It's also possible that the rail could be inductive. The robocar could drive and keep its inductor contact just a short distance from the coils in the rail. This is more expensive than direct contact, and not as efficient, but it's a lot cheaper than burying inductors in the roadbed. It's safe for pedestrians and most impacts, and while a hard impact could expose conductors, a ground fault circuit could interrupt the power. Indeed, because all vehicles on the line will have alternate power, interruption in the event of any current not returning along the return is a reasonable strategy.

For commuters with electric cars, there is a big win. You can get by with far less battery and still go electric. The battery costs a lot of money -- more than enough to justify the cost of installing the connection equipment. And having less battery means less weight, and that's the big win for everybody, as you make the vehicles more efficient when you cut out that weight. Of course, if this lane is only for use by electrified robocars, it becomes a big incentive to get one just to use the special lane.

The power requirements are not small. Cars will want 20kw to go at highway speed, and trucks a lot more. This makes it hard to offer charging as well as operating current, but smaller cars might be able to get a decent charge while driving.


Another solution to the safety problem of electrified rails on the shoulder of a highway would be to have the rail set back in a slot too small for a human hand to fit. A vehicle would need precise driving and a thin flexible contact able to handle the variations in movement.

And on a highway there should not be peds at all. You do need two contacts. Most 3rd rails run 600 to 1000 volts, and use DC (more power for same peak voltage) but at that voltage you need 20 to 33 amps for a car, which is a lot for a small contact. Overhead wires run from 1.5kv (often DC) to 25kv for high speed trains. One problem at those high voltages is running that kind of voltage into the car (to be transformed down usually) presents some risks, I would imagine.

You need to find a sweet spot that is high enough voltage to send the power without lots of current, but not so high as to be dangerous or arc etc.

I'm under the impression that all MBTA subway trains and trolleybuses use 600 VDC, whether they run on overhead wires or third rail. (The Blue Line even uses both, normally with overhead power east of Airport and third rail west of Airport, though I think there's overlap in the power sources from Airport to one of the adjacent stations.)

At highway speeds wouldn't a physical, metal-on-metal running contact with a rail or wire in the shoulder guardrail be rather noisy?

The ambient noise around the undercarriage of a train using a third rail is already rather noisy, so the added noise of third rail systems on trains aren't noticed so much. The train is already going to have significant sound insulation built in to it.

Cars, on the other hand, are designed around rubber on asphalt, and passengers are riding much closer to the wheels, road, and outside of the vehicle.

I could imagine a system that is only or specifically designed for large multi-axle vehicles like trucks and buses would have fewer problems with noise. In that case high overhead wires could be used and the issue of automobiles needing extremely long contact poles wouldn't exist. Of course then it wouldn't be so cost-effective to install the wires in the first place. Unless we go all the way (back) to the idea of road trains that you drive your car up into, for extended trips like between SF and LA...

Would a system like this need to provide full power to run a vehicle without it drawing any power at all from its onboard battery? Since one of the problems of electric vehicles is range, what about having the system simply augment a vehicle's battery, to extend its range? That way you'd get say X miles of range on a charge driving around town, but 2X miles of range with the power augmentation when driving in the "powered" lane on the highway. Your battery is still running down, but only half as quickly.

Without needing to provide full power to a vehicle, inductive power transfer might be more feasible, which would result in less added road noise.

For the Siemens project, they actually are promoting it as being lower noise -- but it is overhead wires, and not highway speeds in the real usage in ports. So yes, a physical contact, scraping and sparking might well be noisier than the gasoline engines removed. And yes, it could be used as you say, just saving some battery but not charging or driving without it.

The operating plans I have seen published for the California High Speed Rail system suggest they're planning on 11 trains per hour on the busiest segment at the busiest time of day, and they expect the track and signal system could handle 20 trains an hour. That seems to assume they'd only be carrying passengers.

Amtrak's Auto Train (Virginia to Florida, automobile required as part of the party's reservation, carless travelers need to take other trains) is one of their more financially successful trains.

This suggests that the California high speed rail system should provide automobile transportation as a high end service. If they look to the Chunnel's version of this service and not Amtrak's to choose a loading method, they should be able to make a high speed train (with the sum of the time spent loading and unloading, time spent at 220 MPH, etc) faster than driving on the Interstate at 65 MPH, for an SF to LA trip.

Isn't it best to lay a rail track and use trains? The old ways are sometimes the best ways. And simpler.

After reading about the South Korean experiments with vehicles running using induction current, I was very excited about the idea of converting most of our vehicles to electric.
I even had some ideas to convert existing freeways to electric at a pretty reasonable cost per mile. The part that killed my idea was when I figured out the power consumption for afternoon rush hour if a significant part of the commuter traffic was electric. Late afternoon is already peak electric consumption, so we would need a huge amount of new power plant construction to provide the necessary generating capacity and the cost was close to a trillion dollars even if we used natural gas, which is cheapest on construction cost per KW. It makes more sense to build more cars using CNG or wait for the battery technology to develop to make it practical to use off peak electricity for power.


While it is important to consider all alternatives, I think electrified rails whether for direct contact or induction, are looking like a dead end.
Battery swap technology looks to me to be the better long term option.

1. The infrastructure of swap out battery stations would surely be many times cheaper than running electrified rails.
2. Battery technology and range are steadily improving.
3. Recharge allows for off peak power usage where as electrified rails would drain the power system at the worst possible time (peak hour).
4. Swapping batteries could be done by an self driving car in between carrying passengers.
5. It is not practical to run electrified rails on all streets so battery recharging infrastructure will be needed anyway.

While it can be argued that electrified rails would extend the range of electric cars by recharging the batteries on selected roads, I still cannot see the benefits being worth the costs, which I think would be enormous.

I have an earlier article on battery swap. Its biggest problem is not so much technical (at least with robocars) but the fact that consumers no longer buy the battery. Only the swap companies buy the batteries, and so there is not a competitive market.

Of course, it's also true that electrification would need a standard and there would be no competition in design of the connection system either.

The reason that people love electrification is both that batteries are expensive (that may be fixed) but also because they weigh a lot. You carry all that weight whether the batteries are full or empty.

Electric cars will need recharging, but enough battery for city operations is already doable. It's handling long commutes and highway driving that batteries can't do. Swap is not so practical for commutes but does solve long highway.

The cost of overhead wires is not so bad. A rail down the side is tolerable too, I think. Buried in the road would be very expensive.


I have to disagree with you that there could not be a competitive market for a battery swap system. As an example consider the humble carbon battery. These are made by a variety of companies in a range of sizes in a very competitive market.Yet they are all built to the same standards so that you do not need to buy a particular brand to match your device. You can buy different brands in the same store which is a separate business to the battery suppliers. I don't see any reason why this basic model cannot be used with a battery swap system. In other words a battery swap station may stock different brands that are of standardize sizes and power outputs. Like gas stations there can be different brands of swap stations. Of course each battery has to be tracked and maintained as well as a re-distribution system to keep the supply balanced to where the demand is. However in the pursuit of profit in a competitive market, businesses are often very inventive.

As to batteries being expensive and weighing a lot, well I can only agree with you. However with slow but steady improvements in the technology I still think batteries would have the edge over electrified rails. Admittedly this is just a guess, it would take detailed costings and study to have a clearer idea.

Of course there can be competing suppliers. But in the battery swap world, the driver does not own their battery. It is owned by the battery swap company or similar and rented by the car owner. So there is only one real customer for the batteries -- the swap company. In the future there might be multiple swap companies but that requires a lot of density because unlike gasoline, it's hard to be able to go to your choice of swap company. You would be dropping off batteries from company A at the swap station of B.

Having only one customer is not as bad for innovation as having only one supplier, but it's still bad.

Tesla seems to have pretty much solved the technical problems involved in driving 500 miles / day with a passenger car that relies on batteries. There are some logistical problems in getting the SuperCharges deployed, and they need to get the cost of the packs down, but they've also got at least a 6 month backlog of 10,000+ customers who've put down $5k+ refundable deposits to hold a place in line (and they're still waiting for government crash testing before they start production).

I do think 18 wheelers are going to have trouble getting a decent range / charging time for long haul trips without battery swap. But once it becomes obvious that personal automobiles have driven the cost of batteries down far enough, I suspect there will be an effort by the trucking industry to have a single battery swap consortium.

I also don't think you'd necessarily need to have a single company owning all the packs. You could instead have a consortium whose customers would be all the battery swap stations and trucking companies, which would lease batteries from battery leasing companies. That consortium could attempt to set high-technical standards and then take the low bidder, much the way government contracting tends to work, but perhaps they would manage to care more about quality than government often does. And if batteries are beating diesel by enough, it won't matter if this isn't quite optimal. claims solar costs are dropping 7% a year and that price drop may be accelerating. (.93 ^ 9 and .93 ^ 10 suggests that this is about 9-10 years for the price to drop to half of what it is now)

If this continues, 10-20 years from now, we may find that charging car batteries primarily during the day is best. (More generally, the prices of both solar panels and batteries keep dropping. The price of local electric monopoly power distribution probably won't drop much if at all. This leaves open the possibility that the typical suburban residence, 10-20 years from now, may find an investment in rooftop solar + batteries to be cheaper over the long run than continuing to pay the monopoly.)

I believe that, prior to the EPA rating of 265 miles, Tesla was claiming that the Model S would do 300 miles at 55 MPH on an 85 kwH battery pack, which is about 15.6 kw, not 20 kw. But perhaps by highway driving you meant 65 MPH, and not Tesla's 55 MPH. (What I really want to see is the 65 MPH range, which I have not seen published. I assume the EPA 265 mile range is based on the assumption that people are going to drive dozens of miles at city speed from their hotel to the highway, which isn't always going to be the case, so 65 MPH range may be worse than that.)

Tesla also claims their 90 kw SuperChargers will replenish 50% of the 85 kwH battery pack in a half hour. This suggests a 500 mile / day road trip ought to be doable with about an hour of charging throughout the day while you stop for lunch, etc, without battery swap or overhead power.

And if the SuperChargers cost $25k each and you want one SuperCharger per 50 miles of Interstate Highway (Wikipedia says there are 47,182 miles of Interstate Highway), about 944 SuperChargers are needed. That works out to about $23.6 million for SuperCharger coverage for the entire US Interstate Highway system, less than a fifth of what was recently spent on a new lift bridge to carry four lanes of automobile traffic across the Chelsea River in Boston. That $23.6 million is also less than a tenth of what is proposed to be spent to rehabilitate the Longfellow Bridge, which carries the MBTA Red Line from Boston to Cambridge, along with automobile traffic.

Of course, if you have infinite money and prefer 800 mile / day road trips, the Model S is still going to be unsatisfying.

Right now this tech does not seem to be ready. Charging that fast reduces battery life in most systems, though it would be great to see a tech where this does not happen.

20kw is a rough number. There are much more efficient vehicles that can do much better too. And the question is, do you want to be able to pull peak power, or use your battery for accelerations? Anyway, as people have noted, almost any decent amount down to the 6kw of a L2 charger is of interest to people so they have more range on less weight.

Note that Tesla is advertising 50% charge in a half hour, not 100% charge in an hour. That's because at some places in the state of charge curve, you can't safely charge at 90 kW. I've seen rumors that for the Tesla pack, you can't quick charge when the battery is either nearly dead or nearly full, and the 40 kWH version of the Model S will not support supercharging at all. (But the 40 kWH version of the Model S is also vaporware to some extent, moreso than the bigger packs: the plan is that they're going to sell an 85 kWH pack to everyone willing to pay for one before they start offering the 60 kWH version, and then they'll sell all the 60 kWH packs people want before they start selling 40 kWH. I will not be shocked if they manage to drop the prices enough as time goes on that they can avoid ever building the smaller packs.)

Meanwhile, I believe the Nissan Leaf quick charge option is advertised as 80% in a half hour; I think a Leaf can be quick charged starting when the battery is dead, but they don't recommend quick charging the battery beyond 80% full. And I believe Nissan is offering a 100,000 mile / 8 year warranty on the battery pack, where if it degrades to the point where it holds less than 60% of what it was designed to hold on the day the car was delivered, and the car has gone less than 100,000 miles and is less than 8 years old, Nissan will replace the battery at no charge. I wouldn't expect a warranty like that if the tech really isn't there.

Mitsubishi's quick charge capability doesn't have these sort of restrictions AFAIK.

I think battery life is one of the things that we get very little data on and I would love to see more.

I am particularly interested in the lifetime watt-hours of a battery at various duty cycles. There will be an "optimum" duty cycle that gives you that maximum lifetime watt-hours and I am curious as to what that is for the various packs and chemistries. I will probably write a blog post on this later. Then I want to calculate lifetime watt-hours per dollar at the optimum duty cycle, and with various use and charging patterns. We want to know how much fast charge affects this compared to level 2 and ordinary level 1 etc. And you want to factor in the cost of the extra weight, because a battery you never discharge to more than 40% carries around a bunch of weight for that excess, which in turn costs energy.

This is interesting because fleets and robocar taxis can pick their cycle for this maximum value, while human drivers will have variable patterns, with some short trips and some long, and some urgent fast charges etc.

There are a lot of variables here -- a lot of usage patterns, a lot of different batteries. But since right now the battery pack is the most expensive part of the electric car this is a hugely important number. If you go 150,000 miles on a $15,000 battery pack, then the pack is costing 10 cents/mile. Since the electricty is more like 2.5 cents/mile, it turns out the pack cost is far more important from an economic standpoint. And of course, in a vehicle like a Prius at 50mpg and $4/gallon, the gas is 8 cents/mile, and the depreciation of the ICE power train is a different number.

If you're in California, you're probably not far from a Tesla showroom, and you could see what they'll tell you. However, Tesla's most recent annual report to shareholders has some tables of Panasonic battery prices, etc, with all of the actual numbers missing, because they wanted confidential treatment. (I haven't tracked down whether the real numbers were publicly released elsewhere, but that certainly tells you something about their attitude towards anything the SEC doesn't force them to share. On the other hand, Tesla's whole annual report feels like it was written by their lawyers; Berkshire Hathaway's annual report is much more pleasant to read.)

The Tesla Roadster didn't begin production until 2008, so Tesla only has about four years of data there. The Roadster also didn't support quick charging, and I believe the 85 kwh version of the Model S uses a newer, higher density battery chemistry than the Roadster. That leaves plenty of unknowns, even for Tesla. reports ``The number of CHAdeMO DC Quick charger installed up to today is 1393. (Japan 1154 Europe 207 Other 32) last update 2012.04.27'' lists 2010 as the formation date. So there probably isn't more than two years of data on quick charging.

(The quick charging standards I'm aware of are CHAdeMO, supported by Nissan and Mitsubishi and perhaps others; the Tesla standard, for which the first vehicles will probably start shipping in the next few weeks; and the SAE standard, which seems to be supported primarily by automakers who are in no rush to build cars using it; some wonder if the whole purpose of the SAE standard is to delay the adoption of quick charging. Tesla is planning to build out a network in the US, and if they're aggressive enough, they could easily become far more widespread than CHAdeMO. Then again, if the long term future involves half the parking spaces at highway rest stops having quick chargers, having two standards might not be a problem.)

I believe some of the car makers do recommend against regular quick charging.

The current prices are based on Tesla and Nissan each building something like 20,000 battery powered vehicles a year, and that 20,000 may be a worldwide figure; there are something like 15 million new cars built or sold in the US every year. It's also not clear that cost of production is a major factor in the short term; if you look at the more desirable parts of New York City or the bay area, much of the cost of a home is outbidding the others who want to live there, and not the cost of constructing the building, and likewise there is limited factory capacity and there are many people who want the cars in the short term.

If we had 2 million battery powered cars being built in the US every year, the need to optimize battery usage might decrease somewhat.

At the same time, there are still some interesting questions as to whether charging daily or twice a day makes sense, for fleet vehicles where you would end up dragging more batteries around for daily charging.

For a package delivery truck, for example, does quick charging during a lunch break make sense, vs dragging the extra batteries around?

For a transit agency optimizing bus usage, the average vehicle can probably spend a few hours in the mid day period at a charging station that isn't especially quick, which might put less wear on the batteries, but there may still be a question of electric rates at different times of day if the batteries become cheap enough, though perhaps it is clear that in the short term, enough batteries to get through the morning peak time plus an extra hour or two, then recharge, then cover the afternoon peak, then charge overnight is the right pattern.

But labor cost can also be an issue if charging takes time. I think I've seen claims that something like 75% of the cost of providing local bus service is basically driver salary, so anything you can do to provide more service with less driver time tends to be a win. That suggests that driving the bus to a charging station not on the route during the day or waiting for a charge during the day may be expensive.

There's also the $15k-$25k for a quick charge station, vs probably sub-$2k for a 240 V at 30-70 A charging station. If every package delivery driver takes their lunch break at the same time, you might not save money buying quick charge stations instead of batteries. doesn't really answer your questions about how batteries will change over the years and over many discharge cycles, but it does have some more information about how driving conditions will affect range; they claim the 85 kWH Model S will go more than 250 miles at 65 MPH, assuming no headwinds (and possibly making some other optimistic assumptions).

It is also interesting that aerodynamic drag seems to be the major factor in performance; this makes me wonder if claims that weight matters may be overstated. Adding a few hundred pounds of cargo presumably has no effect on aerodynamic drag (unless the springs in the car change the angle or height to the ground), although added weight may affect energy loses in the tires a bit, and while adding a few hundred pounds increases the energy required to go from 0 MPH to 65 MPH, regenerative braking ought to reclaim a lot of that when going from 65 MPH back to 0 MPH. claims ``when new the Nissan Leaf battery pack is rated to hold 24 kilowatt-hours of electricity, and that Nissan expects the capacity to gradually decline after an 10 year period to hold only 19.2 kilowatt-hours (80% of 24 kilowatt-hours).'' and goes on to claim that the Nissan warranty is on peak power put out by the battery pack (W), not the total energy the battery pack holds (WH).

I am keen to see if super capacitors live up to there potential. Unlike batteries super caps don't wear out retaining there full capacity after erratic charge cycles and fast charging, they give lots of power and last millions of charge cycles. Whereas current batteries loose most of there capacity after a few charge cycles which makes a Prius a huge money and lithium sink. A short time ago I remember reading about a graphene oxide super capacitor made in a DVD burner which it was said had energy density similar to Lithium batteries. And the next day a read about new way to make graphene oxide cheaply in a ball mill with dry ice and graphite, so it sounds like it could scale up to mass production well and be much cheaper than all the rare earth metals used in current batteries. So that could be the thing that finally makes electric cars economical and attractive. A few fast charge stations would be much less infrastructure to build and tear up, everyone needs to stop for a couple of toilet and meal breaks during a days driving anyway don't they.

If you could actually get a super cap with the ability to hold the amount of energy needed for driving, they would be fantastic.

The Balqon trucks use batteries, as described at and and it seems that putting in charging stations for them is a lot less infrastructure than adding overhead power lines along an entire route.

Overhead power is indeed a huge amount of extra infrastructure.

People are interested in it because it has some very large advantages over batteries:

  • Batteries have limited range, of course. External power's range is limited in a very different way -- to the places where the power is -- but it is not limited specifically by distance
  • Batteries add weight (and cost) to add range, and that weight in turn hurts the performance, as the weight is the same full or empty.
  • Batteries have limited lifetimes and each recharge cycle cuts their life, and the cost per mile can be actually fairly high
  • There are losses in charging batteries and discharging them. Not crushing, but they are there.

Batteries have advantages of course -- the ability to go everywhere and no need for a complex connection system.

Put power underneath the motorway: a linear motor, like with magnetic-levitation trains. Cars then can go long distances on the motorway and need a small battery only for short distances off the motorway.

Anything in the road is hugely expensive. It increases the cost of the road a great deal, and then makes the maintenance hugely expensive later -- you can't just add a layer of pavement. Overhead or along the side are much better.

I assume the 20 kw requirement at highway speeds assumes a similar weight of vehicle to today's cars? Robot cars surely require less than this for reasons you have outlined previously.
There would be much less requirement for hard acceleration if cars accelerated in synchrony as would be possible with robots.
This would also reduce the weight because of a smaller drivetrain.
With far less likelihood of a crash crumple zones and heavy steel could also disappear, again reducing weight and power requirements.
Further reduction in power requirements could come from allowing cars to slipstream, which again only robot cars could do.
If you then assume that robot cars remove the fear factor of travelling in a motor cycle sized vehicle, then 20 kw looks like a lot.

Keep up the fascinating blog, sadly many can't see the massive opportunity that removing human drivers could bring.

A little off topic but..

A British company has announced a breakthrough that could add another layer of navigation positioning for a self drive car. It is a copy of GPS system that uses cell phone towers instead of satellites for signal positioning.
While something like the Google car system of local observation may become the primary means of navigation, having second, third and fourth "opinion" from complementary navigation systems can only increase certainty.

Idea is very good, less polluting, more economical and safer transportation, but we need big investment.Maybe in the future something will be reached, until then it will have to pass some years.

Add new comment