Many of us believe that there's a natural fit between electric drive trains and robocars. It's not required -- you can certainly make robocars driven by gasoline, natural gas, hydrogen or anything else.

Electric has several advantages:

• No emissions at the vehicle, and, as we green the grid, lower total emissions
• Quiet and smooth operation
• Low energy cost (though currently high battery cost.)
• Lots of optimistic news in the pipeline about cheaper batteries, faster charging and better duty cycles
• Small size of power train.
• Most weight in the base for improved stability
• High reliability of electric motors and vast reduction in moving parts for lower maintenance

Oddly, one of electric's big advantages -- high torque -- is great fun for drivers but much less value in a taxi.

Electricity's big disadvantages are limited range, long recharge times and a lack of places to recharge. Robocars don't care about most of those. For long trips where you do care, liquid fuel robocars will take the burden.

For personally driven electric cars, you need charging where you park. And that needs to be very close to where you are going. Once you park, you occupy the charging space, even after you are recharged, even if you needed to park for 8 hours and needed only 2 hours of charging.

Robocars don't need new electrical transmission infrastructure. When they need to charge, they can drive to where the charging is, rather than bringing the electricity to where we drive. The best place to put robocar charging will be in spare space next to power grid substations and high-tension power lines. Here, drawing megawatts is no problem and installing charging stations is a much cheaper proposition. You don't want to drive a long way to get power, but a few miles are not an issue. Cars should be able to query for a nearby available charging station with good prices and reserve it for the time they need, then go there at that time.

Robocars, of course, will only go to the charging station when they can book a slot, and will only stay long enough to get their charge -- they will leave and free up a space as soon as they can. They can also grab small amounts of charge in short bursts during the gaps in their workday.

The upside is that while private electric cars probably need 1.5 charging stations per car (one at home and a fraction of one at the office and other locations) with robocars, private cars need only 0.05 charging stations per car, and robotaxis only about 1/3rd of a charging station per taxi, since they are all put into 100% utilization.

The charging lot

I predict the first charging lots will actually be staffed by a human who plugs and unplugs the cars as they arrive and leave. While it's possible to build robotic arms that can plug in a wire, a busy station would have more than enough cars circulating through it to make productive work for human electricity jockeys. With a wide variety of different plugs and locations, the humans can be very flexible. A fixed robotic station faces the problem that the robot only operates a few times a day, since each visitor comes for an hour or more. (This is unlike a gas station where cars visit for only a couple of minutes.)

It is somewhat ironic that the electric robocar might cause the vanishing job of full service gas station attendant to return. Gasoline powered robocars will want full service stations as well.

While ordinary robocar parking lots will have the cars park densely in "valet" style, because any car that blocks another one in can move when the inner car wants to get out, we can't do quite as much of this in a robocar lot, since while a car is plugged in, it can't move unless the attendant comes to temporarily unplug in. Still, since most cars will know how long they need to be there for charging, it should be possible for the system managing the lot to place cars that will need a long charge behind cars that need a short charge. And in a pinch, a car can indeed be temporarily unplugged to move.

It will also be nice if the lot allows the attendants to quickly just move cords from one car that is ready to leave to another car that's just arrived. The simplest way to do that would be to have two spaces for each charging cord, with the new car pulling into space 2, and the attendant quickly moving the cord from space 1 to the new one, after which the freed car leaves. This uses a lot of space and would only be the strategy when attendant time is more expensive than the land.

It might be possible to make charging equipment cheaper. Today, electric cars contain their charger inside them. The "charger" that they plug into isn't the charger at all, it's just a circuit that tells the charger in the car how much current is available, using an ancient analog protocol. In a robocar, that communication could be done over other digital channels, allowing the "charging stations" to really just be cords with plugs on the end.

Alternately, the chargers could be removed from the cars and put into the ground stations, making them true charging stations. This saves a lot of money -- there are many more cars than charging stations in this world -- but it does not allow the car maker to tune the charger to match their particular vehicle. If a new car comes out that wants to use a new charger, it could do so, but only with a lot of extra cost compared to the no-charger cars. In addition, the charging system for the battery is one of the greatest sources of fire risk, and you may not want to trust another charger to that.

Eventually we might see a mobile robot that can travel the lot, pull out charging cords and plug them in, but not for a while.

Inductive charging and contact plates

There are other ways than plugs to charge. One can place inductive charging units in the floor of the parking lot and charge cars equipped with the corresponding plates. Then the cars can just drive onto the pads. This had advantages but about 10% of the power is lost in inductive charging, which over time can be a significant cost, and the inductive system adds a fair bit of cost to the station and car.

Because robocars can park themselves with perfect precision, you could also use ordinary metal contacts on the ground or springing up from it. After all, subways and trolleys pull tons of power by this method. All this takes is spring-loaded contacts and a single actuator to deploy and retract them. These could be in the vehicle, or even in a box on the ground with fairly low cost. The car would negotiate digitally when it is positioned correctly on the box, and test the contact quality with lower current to start.

A retractable cover plate to keep contacts clean also makes sense, and can be retracted with the same actuator. The plates on the ground don't need to be flush with the pavement, they can be a box sitting on it, as long as only robots drive in the area.

The biggest concern is again flexibility. If you change your mind about what charging method you like, it's more difficult to change ones that have large physical components. It is possible for a charging lot to support different types reasonably, however.

Lots can have a mix of inductive plates, human-plugged cords and other methods. A single spot can of course be served by both an inductive plate and regular cord if that makes sense.

Battery swap

Battery swap has failed for human driven cars. It's expensive -- expensive batteries, expensive swap stations. Cars need to be able to experiment with different battery designs, so the swap station has to stock and handle tons of battery designs. The swap takes 5 minutes, so if there are several people all wanting a swap at once, the wait is not acceptable.

No problem for robots: They can schedule their swap, and they don't mind waiting. They don't mind going to a particular place that has their style of battery, fully charged as long as it's not too far, so you don't have to support all batteries at all stations.

The only issue is the cost. Swap puts a drained car back into service as soon as it can schedule a swap. You want that for your full time taxis at peak hours. The answer depends on what the cost of batteries is compared to the cost of the car. In many cases, it bay be simpler and cheaper to make more cars rather than just more batteries which can be swapped. Having more cars has many advantages in a fleet, particularly at peak time. You have to have enough cars for peak demand no matter what. As you get to off peak time, some cars start going out of service due to lack of demand, and they can then charge the slow way.

You thus only need swap for cars with small batteries -- batteries so small that they can't stay in service for all of the peak and near-peak. Swap might also be useful for long distance electric cars, but in many cases it's easier to swap cars than to swap batteries, unless the passengers have a ton of gear with them. Or use liquid fuel (including green biofuel) on long haul trips.

Geometry of the lot

In medium density areas, it makes sense to install charging next to transformer substations. In denser areas, it will make sense to bring in power to existing parking structures, since there will be a glut of those. In the unlikely event of building a new structure, it would look a bit different from today's lots. Ceiling height would probably be quite low on all but the first few floors, just high enough for the attendant to walk.

When to charge

As we get more electric cars, we will start to need more generation capacity. US electrical consumption is about 4,100 TWH per year at present, with 1/3 each going to residential and commercial and industrial/other. Americans drive 3 trillion miles per year, about 68% urban and 31% rural. If all 2T of urban miles were electric, that's a fair bit of electricity. Electric cars today take about 240 wh/mile, but single person electric cars can easily do under 100. Assuming a 50-50 split, that's about 350 TWH of electricity needed, which is less than a 10% increase in electricity demand.

Most cars will want to be in use at rush hour, and so would charge either in the evening, or mid-day. Evening power is a good way to sop up the baseload from nuclear plants and other capacity -- generally capacity is going spare at that time of day.

Peak demand comes during the afternoon, but the day is when solar power comes online. The massive drop in the cost of solar -- it is now, in sunny places, the cheapest type of new power plant per kwh and it's getting cheaper -- bodes well for this. The big problem with solar and wind is the unpredictability of their supply, but charging batteries is the ideal use for such sources of energy. Cars will charge when there is spare power, and to some extent, they can just not charge during the periods of peak demand. These periods are generally predictable and are based on temperature.

Robotaxis will have to decide how large a battery they need. In some ways, smaller is better because it weighs less and costs less, but you must make more trips to a charging lot. A larger battery will give you not just the ability to do long trips but to delay charging in order to charge when prices are lowest. Batteries will be sized for the sweet spot. If your taxi only has a little power, you just only send it on short jobs you know it can handle. Even if your personal car runs low, you can just get out and switch to a taxi while it scurries off to some charging it booked.