Brad Templeton Home


Robocars
Main Page


Brad Ideas
(My Blog)


Robocar Blog


The Case

Roadmap

Roadblocks

Car Design

Problems

Privacy

Stories

Deliverbots

Whistlecars

End of Transit

Objections

Myths

Geeks

Parking

PRT

Accidents

Notes

Teams

Government

Motornet

Urban Planning

Development

Congestion

Definition

Statistics

Glossary


Sidebars: Charging

Speed Limits

NHTSA Levels

Valley of Danger

Simulator

Google Cars

A Robocar Airport

A Robocar Airport

Modern big city airports have become behemoths, often costing billions of dollars. The transition through the airport has become the bane of all who travel, and for air trips under 1,000 miles it is routine to spend more time with ground logistics than actually flying.

SFO has 4 terminals (1 not in use) and 7 concourses for 30M pax/year
One airport I use in San Jose just tore down its last old terminal for a gleaming new billion dollar "modern" terminal. The old terminal was at ground level, and used roll up staircases. Recently it switched to roll up ramps, presumably for the disabled and those with rolling luggage. I loved this terminal. The time from the plane landing to getting into your cab or car was far shorter than at the new one. The walks were much shorter as well.

Early, pre-robocar technology might allow the design of a very different airport. I say pre-robocar because it doesn't require the "any road" vehicles that I call robocars. In the controlled environment it is more accurately termed wheeled PRT.

Heathrow airport has almost completed a simple, pre-robocar PRT system for their 8 billion dollar terminal 5. This technology is known as ULTra PRT and will take people from terminals to the parking lots in small, on-demand, autonomous cars that run down a dedicated track. The track is just pavement with small concrete curbs along the side. But what if we designed an airport based entirely on such technology? Let's consider a fictional trip through such an airport, built using robocar technology available now or in the very near future, which is able to operate safely on controlled pavement without civilian traffic.


Lisa, a passenger, begins her trip by checking in for the flight on her mobile phone. A "boarding pass" is sent back to her phone, which can be put on the screen and displayed to scanners, or exchanged over bluetooth. She checks in for her flight departing at 2:30 pm. The boarding pass tells her to be at the airport at 2:10 pm -- 40 minutes later than she typically has to get to a parking lot at older airports.

She pulls into the airport parking lot just before 2:10. As she hunts for her parking spot, a small robotic car follows her car, and is waiting for her when she gets out. Her phone exchanges the boarding pass via bluetooth and it opens for her, and she gets in and presses start. The airport is a sparse place, mostly the parking lots, airstrips, a modest number of low buildings and hangars, and a series of jetbridges spaced along the taxiways.

The car heads out of the parking lot to a private track for these robotic cars. It moves along the track to one of the small security buildings on the tarmac and docks there. While it was moving she pulled out some trays stored in the car and put her purse and laptop in them. At 2:14 she quickly gets out and finds she does not have to wait in line for security, because she has an appointment which was arranged by her airline for a small extra fee. Most people get these appointments, though some who don't have them or missed theirs are waiting in a line. While there are several security stations, the robocars arrive at them to balance the load and keep all screeners occupied.

She shows her pass again and the security theatre begins. By 2:18 she's through security, and she waves her phone and walks through an electronic gate to another waiting robotic car. This car zips right out onto the airport tarmac and moves directly to her plane. The plane is parked out on the tarmac near a "jetbridge" which has connected to the front door of the plane. The jetbridge is not connected to a building, it just has a tower which may have stairs, a spiral ramp, an elevator, or a pair of car-sized ramps to allow cars to drive up, dock and come back down.

Lisa's car actually pauses. Based on her seat assignment, the airline has decided she should be the 30th person to board from the rear of the aircraft. At the rear a set of steps -- also robotic -- has rolled up to the rear door. The steps are covered and the base is connected to another robotic vehicle, this time a long rectangular building about 8' by 50' with many docking ports for the robocars. Other robocars are docked and disgorging passengers; her car brings her in to be the 30th person to dock at the room. Other cars are going to the jetbridge's docks, and passengers are going up the stairs there or using its elevator. All the people in wheelchairs or with strollers used that entrance and are boarding from the front. At 2:25 she is in the plane and gets to her seat. She stows her bag and straps in and waits a few minutes for the last few passengers.

The airplane had stopped at a jetbridge close to the downwind end of the runway, and so it begins a short taxi at 2:30 and has taken off within just a few minutes. After her hour long flight, the plane lands at her destination, and moves to another station with a jetbridge, but this time at the upwind end of the runway, so there is just a short taxi again. As the plane parks, the jetbridge approaches the front door, and robotic steps approach the rear door. She goes down those steps to see a fleet of robocars waiting at the base. The weather's sunny here so the docking building did not get used. She picks one and flashes her pass at it. It takes her quickly to the border between the secured and unsecured zones, where it docks across the platform from another large array of robocars. For security reasons, cars always stay within the secured or unsecured zones, so she has to briefly cross the platform and go through a one-way gate to another car.

In the meantime a scissor-lift truck has gone to one of the other doors on the plane she landed in, and cleaning crew and catering crew in it are servicing the inside of the plane. A special luggage ramp has also rolled up to the cargo hold -- more on that later. A fuel truck has also rolled up and is refueling the aircraft.

She flashes her pass, and it knows she has a Hertz car waiting. The car takes her directly to Hertz, which has been informed she has landed and has her car ready. Lisa knows that in not too many years, robocars will be able to operate on the city streets, and so she will be able to get directly into her robotic taxi right at the security boundary, or in some cases right at the foot of the steps.

Lisa's trip back is different because she's taking back a large parcel as checked luggage. Now she has to arrive 10 minutes earlier for her flight. The only reason for that is to give the security crew time to inspect her luggage. (She wishes she were on one of the planes that has bomb-proof deforming cargo containers and allowed light packages to be checked later.)

At the rental car return, she goes to a station that prints a luggage barcode and RFID which she applies to the handle of her bag. She then places the bag into a small cargo robot. The robot measures the weight of the bag, and since it's OK, zooms away to security baggage inspection. She also gets into a different robocar which takes her to the security building for her security appointment, just like before.

The rest of her trip mirrors the outbound leg with one difference for Lisa. Her robocar takes her right to her regular car in the parking lot, but now she gets in it and briefly waits. On her phone, she can see the luggage robot making its way to her car. In a few minutes it arrives, she grabs her bag and is on her way.

When Lisa's plane landed, a special ramp rolled up and extended into the cargo hold of the plane. The ramp was wide enough for two luggage robots. As soon as it was in the hold, luggage robots rolled up, bearing bags for the next flight of that plane. For now, a luggage handler who had climbed into the hold took out the incoming bag, and placed a bag like Lisa's into the robot. The robot read the RFID and went down the other ramp.

Due to security, this robot could not go directly to Lisa's car, because it might be tampered with in the parking lot. So it went to the security station where it met another robot and transferred the bag to it. Some bags were odd sizes and automated transfer is not used -- a baggage handler moves the bag to the new robot. Once in the new robot, it scurried off to its destination -- like Lisa's car in the parking lot, though it could have been any number of nearby places.

In the future, when the robots can travel the streets, the luggage robot will go directly to Lisa's house or office as she wishes, so she won't have to wait for her bag.

When Lisa was flying, she looked down at this new airport, and saw how different it was from the airports of the year 2010. It was mainly just a runway with the jetbridge stations. There were a few buildings close to the runway, such as the security buildings, and a few buildings further away for airline employees and hangars for working on aircraft. Missing are the giant terminal buildings and long concourses full of gates. In their place is a more modest sized building with shops, restaurants and other services that Lisa didn't use because her trip through the airport was so short. Sometimes she goes to these buildings when she has a long connection. For short connections, she just sits, reading or working in the robocar that takes her from one plane to the other.

The airports of the past were giant edifices that cost billions of dollars. They were grand spaces, showcases of architecture. They tried to make your stay in the airport as pleasant as possible. But people realized they didn't want to stay in airports -- they just wanted to get through them as quickly as possible.

The new airport cost just a few hundred million dollars, less than a fifth of the old designs. Of course, many cities still have the old airports and use them for many flights. The new robocar stations have been placed on the other side of the airport, on vacant fences or where private aviation facilities are. Some flights use those, and some use the big terminals, but there is a problem, as the airlines that offer the robocar service are more popular. On top of that, they are cheaper, as they don't have to pay large fees to use the expensive multi-billion dollar facilities. In some towns, some of the old terminal facilities are being re-purposed into shipping depots, airport hotels, private aviation FBOs and plain old commercial offices.

Design variations

This is a basic mockup. Needs ramps, covers, luggage-bots etc.
In this airport, planes come to "dock" at stations that will be spread along the taxiways. Buildings are not needed at these stations, except perhaps small ones for workers.

For the front left door of the aircraft (usually the only door at a typical airport) it might be simplest to use existing "jetbridge" technology. The jetbridge is anchored not to a terminal but a small tower with some form of access to the top from the ground -- staircase, ramp, elevator or robocar ramp. Aircraft would park near the jetbridge and it would dock as it does today, and someday will be automatic. This would be used by passengers boarding from the front, including all those who can't handle stairs.

As an alternate to the jetbridge design, a cheaper system might involve a similarly adjustable robocar track and special robocars that have doors in the front and back. These cars would go up the ramp to the aircraft door and connect in a line. As the doors open they would form a long tube. Fold down seats would come up and the line of robocar would become a narrow walkway. As passengers moved forwards the last cars in the line would empty and pull away, allowing new cars to come in and let their passengers walk to the plane.

Also much cheaper would be a system with no jetbridge, and just a fancy roll up enclosed ramp with a heated docking port room. Though human drivers would have problems quickly moving such long structures, robots could readily do it. For the highest aircraft such as the upper deck doors of an Airbus 380, it might be necessary to have a more permanent and higher structure. (On the other hand, this is much less than the current burden of adapting airport terminals to the giant A380 since otherwise all that is needed is strong open tarmac.)

As another plus, with no jetbridge, the plane can depart its docking station by going forward, once the ramp wheels out of the way. No need for towing or backwards steering -- faster departures and faster turnaround if space is available.

For most planes, passengers would also use rear and middle doors. These are barely used at most exports except on big planes doing long flights, and often not even then. While using doors on the other side doesn't help much because the aisles are the bottleneck, using back and middle doors can double or triple the speed of boarding.

Wheelchair compatible ramps are compact and cheap today
These doors would use slightly fancy drive-up staircases or ramps, enclosed for weather. The staircases would be robotic but could also be driven by workers. The staircase would include a long piece of "corridor" with docking ports for robocars. (A long vehicle like this is easy for a robot to drive, possibly with control of multiple steering points or motors.) The corridor would be long enough to bring in people as fast as the aisle bottleneck allows. Instead of a staircase, there could also be a roll-up ramp. Today, due to the needs of the disabled, there are switchback ramps that can be rolled up to planes. In the open space of the robocar airport, there is room to not even require switchbacks as planes can be placed far apart.

(Ramps help the disabled of course, but are also nicer than stairs for people with roll-aboard bags. If stairs are used, those who don't like them could also use scissor-lift elevators, which might go to a door on the other side if one is spare.)

It would also be possible to have a small gate building with waiting area, but this defeats our purpose. People don't want to wait. They can wait easily in the cars themselves, and as long as bathrooms are within 1km, then 50kph robocars can have you there in a minute.

Entering the system

Key to the system would be robotic transportation for both the passenger and luggage from the point of arrival -- be it an off-airport parking lot, passenger drop-off, airport hotel, rental car return, transit station, taxi stand or otherwise. This would move passengers of all abilities quickly and without need for buildings.

Today, there is less and less need to interact with airline employees. Passengers without checked bags rarely talk to one, and even the level of interaction for checking bags is reducing. For dealing with problems with tickets and reservations, a video-conferencing station inside the robocar, equipped with hardware to scan and print printing documents can meet the vast majority of needs. For other special needs a small building can exist on the system with airline employees for in-person discussions.

With a robot system for bags, passengers can check their bags right where they enter. The baggage robots can have a scale to measure the weight of the bag. If it's overweight a video-conference can be set up if there is anything to discuss, but in most cases today that just means paying. The baggage robot or the station can have a machine to produce baggage tags, which might be RFIDs or bar codes. The robot will have scanned the passenger's boarding document, and will be able to scan the applied luggage tag to assure it was correctly attached. If not, the robot can go to an airport employee to fix the problem.

People may wish to just affix permanent bar codes to their bags as a backup to the destination coded tags. (Or rather, the destination coded tags might be considered a backup for the permanent tags.) I would be worried about the privacy issues on this if there weren't already 100 things tracking us as we fly.

People could also bring their bag in their own passenger robocar and hand it over outside the security station if that works better.

People who arrive without having done online check-in would have to use a check-in kiosk at their entry point, or ask to be taken to a room where this can be done. Pretty soon very few people would want this sort of delay. It might even be possible to do this inside the robocar on the way to security.

Indeed, the primary destination for most people will be security, and ideally it will be at a time for which they have arranged an appointment with security. Today we are told when our plane will depart and we have to guess when to arrive, not knowing if there will be a long security line, or a crowded airport. We don't really want to know when the plane leaves, we want to know when we should be at the airport. The robocar's trip from an arrival point to security can be well predicted, and security appointments can thus tell us when to be at the parking lot, and that's what we want to know.

After security when people are boarding the car that will take them to the plane is a good place to do the boarding control. Right now that's done by gate agents who scan boarding passes and in rare instances also check IDs. There are a number of other places to do this, and in many of them it can be automatic.

If people travel through an isolated "gate" of modest length, so that it's easy for computers using 3-D cameras and scales, combined with staff doing visual observation of the whole area to assure that only one person is going through at a time, you can make sure you have one person per boarding pass. The very large robocar boarding station after security would consist of a large number of such lanes leading to a single robocar boarding station. People would move through the lane, get their pass scanned and go into the car which will then know their destination. Groups would go through one at a time but the system would see they were all together and let them all through to the car.

The simplest approach is for the cars to go to small docking stations at the plane. You don't dock until it's your turn to get in the short line on the ramp or steps to board. If it's not your time, you wait, seated, in your car. At the plane, the bottleneck is always people standing in the aisle putting luggage away, so there should be no problem bringing in cars fast enough with enough docking stations.

In the case of modest waits (flight delays) cars would move to waiting areas with power to recharge their batteries and provide power for the climate control. For long delays they would travel to a waiting building with shops and restaurants -- the last one to remind you of the old airport designs.

We could also do this in a more traditional way, with a full sized waiting area in a small building by the jetbridge. But it's hard to think of a reason.

At the security station, you would have to get out and go through whatever security show is in vogue at the time. The robocar airport helps a bit here but can't do much about this problem. We can, however, give people appointment times to go through security, and deliver them to the security station at just that time, making only those without appointments wait in line. The security station can also be much more spread out, leaving plenty of table space for people to do the security dance.

The luggage

Luggage travels in small robots meant for small numbers of bags to security, where it arrives in batches, pre-sorted by flight, and within each flight by final destination. This will allow bags to go in to sections of the hold: bags that will stay on the plane at any given destination, bags that will be taken off for claim and bags that will go to different connections, groupd together, sorted by how short those connections are. The ordering won't be perfect since some bags will arrive early, some late and some will take extra time, but loading and unloading should be much faster.

Groups of bags can then go larger luggage robots, and stragglers into small ones, all of which head off to the plane. At the plane we might start with the traditional conveyer belt, but not too far in the future, robots might do some of the loading. If a ramp goes into the hold that can be driven up by a luggage robot, it can go right there to drop off or receive a bag.

For arriving planes, the mostly sorted luggage will either go into a larger robot bound for security if it is to be claimed, or into smaller luggage robots which take it directly to connecting flights.

The luggage to be claimed goes to security where it is transferred to non-secured individual luggage robots. This is unfortunately necessary to keep robots from moving from the non-secured area to the secured area. The individual robots will then take the bag to the passenger's ground transportation point or other airport facility. It would be faster and cheaper if the robots could go directly to the passenger (and in fact in this case the robot might well beat the passenger in getting there) but for now there's no way to guarantee such a robot has not been tampered with.

The robots will be able to go to local destinations like airport hotels. In the future, a road capable robot might take the luggage right to the passenger's home. Before then, most luggage claim areas can be put to other uses.

Landing passengers

Arriving is even simpler. If there are doors other than the main one, a large number of personal robocars can come on the pavement to collect the rush of people coming down the stairs. For the jetbridge door, use of the stairs or ramp going down would be recommended to avoid needing a very high capacity elevator. The "robocars to the door" approach has a harder time clearing that door's outflow quickly.

It may also make sense to use larger bus-type robocars since almost everybody will be going back to security to transfer to a non-secured personal vehicle. Vehicles that perhaps hold 10 might make sense so they're not too big to be used efficiently for groups of flyers who are departing.

While on the way to your ground destination, you could probably call up a mobile phone app to see your luggage moving through the system.

Passengers making a connection will take a personal car that will go directly to their next plane, unless the connection is long. This will actually allow very short connections -- that is if the planes are on time.

Other Buildings

There will be other areas of the airport also reachable by the secured and non-secured robocars. Because of the car system, they can be more remote if that's desired, and they don't need to be connected.

Airports will still need bathrooms naturally, not too far from any place that passengers wait. There will also be one or more buildings which have some of the traditional facilities of an airport concourse, like shops, restaurants and entertainment. People with long connections will want to use them. In some airports there would be an international transit lounge, but many passengers might just wait in a a car.

There will be special needs offices for the airlines for those who have to talk to a gate agent to handle special luggage needs, flight changes and the like. However, ideally airlines will also allow most of these transactions to be handled by remote employees using video terminals, which should be found around the airport and in all the vehicles. These video terminals would feature both a reader that can read existing documents and print new documents or other tokens for passengers. Seeing an agent in person need only be done if the agent has to personally inspect something like a passport or piece of luggage. (Even then, only security staff seem to want to inspect ID really closely for authenticity. Most airline staff just need to confirm you have it.)

There will also be buildings for employees to work, airline lounges, cargo facilities and other operations scattered wherever it makes sense to put them. This includes currently unused space at the edges of an airport. For newly constructed airports, it makes sense to put tunnels under the runways before paving them, to make access to far sides of airports easy for the cars.

Airport hotels and car rental companies would of course want to join the network. The cost is just a small strip of paved track with curbs and possible safety walls. Because only the small personal cars would run on these tracks, they could also be elevated inexpensively since weight bearing requirements are low. Other businesses and office parks near the airport might well want to join the network.

If airport hotels are on the network, they can provide fine restaurants, meeting rooms and short-term sleeping and shower facilities, though generally this would mean being outside the secured zone.

Security

This airport would be even faster and simpler without the security issues seen at today's airports. The design above features two complete systems of cars and luggage robots. One system stays in the secured area, the other operates outside it.

While cars and luggage robots could readily move from planes out to non-secured destinations like hotels and parking lots, it might be possible to tamper with them there, hiding weapons which would then travel back into the secured area inside the robot, possibly to be extracted once there. In most cases this would require an insider (since luggage robots don't have contact with passengers in the secured area) but the risk is higher for passenger cars.

Having all exiting cars return to a security building so the person or bag can be moved to another vehicle has another advantage -- the robots have to go there anyways, to pick up passengers and bags that have cleared security and need to get to planes. With direct drop-off, the vehicles would need to travel to the passenger's destination, and then go to security, though possibly carrying an incoming bag or passenger.

It's not out of the question that luggage robots could be used on both sides of the boundary. The luggage robots can be very, very simple, with plexiglass fames, and the bags can be x-rayed, sniffed and inspected while on the robot. This means the robot is also inspected and thus can re-enter the secure area. The hard-to-inspect areas (The motor/battery unit) could possibly be made very solid and tamper-proof, with tamper detection circuits. Since the luggage robots only encounter luggage handlers in the secure zone, this could be secure enough -- after all, corrupt insiders have other way to smuggle things in besides ripping open tamper-proof robot casings.

The big advantage of luggage robots that can move quickly is that they would often reach the passenger's destination just before the passenger -- no waiting. This would make checking luggage much more acceptable, even preferable, to some passengers. That in turn means less chaos in boarding, fewer problems with overstuffed overhead compartments and faster plane turnarounds. The only hassle would be needing to arrive earlier to allow time for bag X-rays. Live tracking of bag progress on smartphones would keep passengers happy.

Bomb-proof luggage containers and holds might be one answer to this. I don't know the status of products in this area; they seemed promising at the start of the century. In the future, when deliverbots are allowed on ordinary streets, all the hassles of checked luggage will go away -- you'll send your bag ahead to the airport, and it will be delivered right to your house or hotel unless you want otherwise.

Passenger cars could take people directly from planes to some destinations if there are secured paths built to those destinations, using things like enclosed or secured elevated guideways. Some of the locations can be placed around the edges of the secured area -- car rental depots, taxi stands, some airport hotels, transit stations and more could be placed so that the cars drop passengers off at a one-way hallway that goes out of the secure area but doesn't allow anybody in. It's cheaper if that can be automated and doesn't require a full time security staffer the way it does now.

Of course, a shared car rental center and some other destinations could also simply have their own security station. The volume of passengers going to and from a rental car center is probably enough to justify that. This would mean you could drop off your rental car and walk through security to get into the robocar that will take you right to your plane.

Changes in security stations

Today's security stations are usually crowded, particularly if they were not rebuilt after 9/11. While an airport might have many security stations, you generally want to use the one close to your flight. Space is at a premium in airport terminals.

The best thing done to speed up security right now is putting out more table space so people can do all the unpacking they need to do in parallel. The more space you have for tables the more you can move the bottleneck to the X-ray machine, and then you decide how many of those you can afford.

In the robocar airport, security stations don't have to be that close to your flight and you can use any one -- in particular the one with the shortest lines. Passengers can be delivered to whatever station is optimal, at whatever time is optimal -- not necessarily in the order they arrived, though that may be fairest. Passengers can get appointments, or pay extra for priority when they are late for a flight. (The airline can pay extra for that too, as they often do for elite and 1st class passengers.) This payment doesn't slow the line down for the non-payers, rather it pays for extra screening capacity that speeds up everybody.

The individual security lines can also be spread far apart. This has one very positive consequence, as there are not giant crowds at security. Each line can be short, with cars bringing in new passengers as needed. Today security lines have a real security problem -- by definition they are unsecured areas and they have large crowds. A terrorist with a roll-aboard bomb can do terrible things before even getting to security. In a robocar airport, that target is no longer worth hitting.

If one does boarding pass verification as passengers exit security and board the car to their plane, the staff workload can be reduced. However, it does mean that passengers who decide to not go directly to the plane (and instead go to restaurants, shops or the bathroom) would need to re-do their boarding pass check.

It's not out of the question that somebody could smuggle a child into a roll-aboard once in a car, though how they get the child out on the plane is hard to say. (The idea is that being able to do things after the boarding pass check lets you switch flights, thus getting a child onto an expensive overseas flight using a local ticket.)

There are a few other options for boarding pass verification. For example, plane seats could easily have sensors that tell if somebody is sitting in them (as all car seatbelts do today,) and boarding pass scanners right in the in-flight-entertainment electronics. (It's just a cheap webcam.) The plane could tell if anybody is on board who should not be, as long as FAs do a walk-through to look for any cheating attempt. People can switch seats, the counting algorithm need not demand you are in your assigned seat. If a plane does not have this equipment it would need an alternate method, including just having people scan as they enter the doors, where there is usually a bottleneck anyway. In the robocar system it's pretty difficult to get to the door if you are not on that flight, or to get there much out of the order you should board, so there will be few altercations.

If after-the-fact enforcement is sufficient, the systems can also tell if somebody enters the secure area and doesn't get on a plane.

To do it after security, there would be a large station where people boarded robocars to their planes. This station would have a lane to each robocar. At the lane you would show your boarding pass and walk through. Security scanners and guards would assure no more than one person walked through the lane at a time to match boarding passes with destinations. These would be similar to many of today's subway entry lines with mechanical gates which open for a passenger who has put in the right ticket or card.

(Note that all of this is revenue based security, to stop people from buying a cheap ticket and sneaking onto a long-haul flight, or sneaking frequent flyer miles. The security against attackers is all done at the security stations.)

There are a few physical security issues. Due to emergencies, people might need to exit a car while on the tarmac. This would be allowed but would set off alarms and transmit video. (This is already the case as there are doors from the terminals to the tarmac that sound alarms if you use them.) At the actual planes, you need to be sure passengers can't wander off, or wander near dangerous things. The use of docking buildings or at least small docking perimeters can solve this. Robots could actually quickly erect a fence (physical and laser) around a stairway and docking area to prevent this in nice weather, but in bad weather people will like being indoors.

(Of course, the whole security system is a bit of a farce, but for now we are constrained to achieve its goals, whatever they are. And I'm not in favour of all this ID checking anyway. I think a better approach to that would be to have the purchaser of a ticket provide a photo of the passenger, and to compare that as they board. Their name is not required.)

The goal is to be able to run the process with very few employees, so that the employees you do have can be working on special cases, and not routine boarding and handling of passengers. However, in the end it may not be that expensive to have them do it, as they do it today. The airport building is still vastly cheaper.

Some interesting variants

If an aircraft can be accessed by both front and rear doors it becomes possible to both empty and load an aircraft at the same time, with passengers exiting at one end and coming in on the other. This could mean a radical reduction in turnaround time, which is a big cost center for airlines. Cleanup would be the responsibility of passengers, but this is not that radical -- trains work this way.

Alternately it could be possible to have passengers exit at one gate and enter from another. Consider a north-south runway with aircraft taking off and landing from south to north. Arriving short-haul planes would immediately taxi to a nearby "gate" at the north and passengers and luggage would be removed, and the plane fully or partially serviced. Once all passengers were off, the aircraft would taxi (or be towed by robotic tug) to another gate at the south end of the airport. Once passengers and luggage were aboard (the 2nd half of any service being done during this interval) the plane would immediately taxi to the south of the runway and take off.

The win here is on short-haul flights. While turnaround is slightly longer, the passengers do not face the often fairly long wait between touchdown and actually getting to the distant gate, and the corresponding long taxi to takeoff position. These often take 5 minutes or more and could be reduced to 1, or a savings of 8 minutes on short-haul flights which may have spent only 45 minutes in the air. It's something people would pay for. Some service could even be done during the transit between gates, such as cleaning and some restocking.

Passengers care most about how long it is between when they have to arrive at the airport for their flight until they are on their way out of the destination airport. These robotic systems can cut that time down greatly. Airlines care about turnaround time, and so their goals and the passengers' are mostly the same. Only the system where the plane moves to another "gate" between letting off passengers and taking on new ones presents a trade-off between passenger speed and turnaround.

Larger cars and Peak Loads

For premium service, solo passengers would like to go in solo cars. This also allows the system to have the cars bring specific people to planes or security at specific times to speed boarding and processing. Solo cars provide privacy for working and talking on the phone, and usually they mean no waiting.

You need more cars if people will be solo, and that costs money, particularly in early development when cars cost more. The alternative is to have mostly multi-person cars and to try to group solo travelers and even pairs into those cars. Most airports will need to do this when they are at their very peak periods, as you probably won't buy enough cars to give each solo flyer a private car at the peak.

This is certainly doable for the trips from major airport entry/exits, especially at peak times. At peak times, the wait for another passenger in a parking lot or rental car center will be measured in seconds, not minutes. At transit stations using larger cars to security may be the only way to clear out an arriving train quickly.

For trips from security or concourse buildings to planes, it will not be hard to group people. To avoid long waits in cars when demand is high, a post-security waiting concourse may make sense. Here, contrary to today's patterns, you don't want people to arrive very long before their flight, if that means they will wait in a car for long periods, tying it up. The arrival pattern for passengers will no doubt follow some sort of bell-like curve, and for passengers within a few deviations from the mean, grouping them will be no problem. Outliers may need single cars, particularly late outliers who do not have time to wait.

If two solo passengers share a car, you either will need to have a place for the one who should board later to wait, or have some out-of-order boarding at these peak times. If passengers should use different doors (ie. one is seated at the back of the plane and the other at the front) their car can dock at both ramps, going first to the door that will be used first. For planes with open seating, the later-boarding passenger will just get the perk of boarding earlier.

In many cases the passengers will actually wait in their own cars, if they are parking a car or returning a rental, or at hotels. This is quite counter to the usual "get to the airport early then get to the gate fast" strategy of flying today. Each airplane's passenger control system will be receiving a feed of where all its passengers are, and it will know what resources are available to move that passenger. Any passenger told, for example, to wait in their car or the lounge building because they got to the airport too early will be allocated priority access to security screening and transit cars when the time comes for them to be delivered to the plane. With reliable trip times, this can all be plotted in advance, and the pool of cars can be reserved in advance. The airline, knowing all the passenger patterns, will reserve sufficient cars to get the job done on time, or face a late departure. Airlines might even own some cars of their own to make sure they can meet some needs. A market approach, where the airlines pay per minute spot rates for car reservations and usage would allow the private parties to work out the most efficient use of the vehicles.

Downsides and counter-forces

This airport is vastly cheaper than today's new airport designs, as well as being faster, more reliable and more secure. So of course the people making the billions on today's airports will oppose it. Some airports think they are there to extract money from passengers, with overpriced food and shops.

Many cities build their multi-billion dollar airports as showpieces. One can understand this, it is the first impression of many visitors to the city. It will take some getting used to that people are just driven across the tarmac to a rental car depot rather than exiting though a grand hall with a sweeping roof. Better to spend the billions on something else interesting that attracts visitors to the city.

Few cities build brand new airports. This approach would probably first be used to expand existing airports, instead of building a new terminal, for example. I suspect passengers at the old terminal might get jealous of those using the simple robocar based flights. Cars must be allocated first to the passengers who need them, but there's no reason that spare robocars could not move passengers to the old terminals, or from security to the old gates, especially for passengers who can't handle the long walks. If the airport buys more cars, this can be offered regularly.

Some people will surely lament that the city has a billion dollar font door and everybody wants to take a robocar to go in the back door.

As cheap as this design is, there will be pressures to make it even cheaper. One way to do that is to try to have more multi-person cars. This can be done by adding a small amount of waiting and coordination to passengers. For example, after leaving security, the system can tell if there are several unrelated passengers all going to the same plane, and ask them to use a multi-person car. Unless you're lucky, this does muck up the perfect boarding order system.

In snowy climates, the tracks for the cars must be plowed and quickly. The robots will have to go a bit slower on snow and ice. Since the robocars are small, it's not out of the question to consider having coverings and enclosures around the car tracks -- this means you only have to plow the intersections. Of course, robot plows would be part of the package, and some of the passenger robocars could actually be heavier duty snow units that can attach plows.

If a vehicle breaks down, just send a replacement, and watch the passengers transfer on a security camera. Then send a crew to go remove the broken down vehicle. One reason you want to avoid fully enclosing the track is so that breakdowns can be just driven around.

There are some centralized elements here, in particular the system which schedules people to arrive at the exact time they should arrive. If this breaks down, the cars will still be able to go to their destinations, but there will be a bit more chaos as people arrive in random orders. The airport will still work, as this is what happens today -- but we will get quite used to the just-in-time delivery of passengers and thus suffer if it stops. Otherwise operation of the vehicles should be mostly autonomous so there is no risk of system-wide breakdown.

Most of the plans here have distributed controls. Each plane would independently be summoning cars to dock near it. The security system would allocate passengers to security stations but after that it would be manged by the car and the station.

In a major storm, it is possible to imagine the cars getting stuck and eventually running out of heat. As such, emergency buses which can rescue anybody stranded by such an event are needed. (In theory this would only be those surprised by the storm, since such a bad storm would be causing flight delays and people would not be going out in cars.)

The use of simple roll-up ramps is not quite as comfortable and glamourous as permanent heated jetbridge structures. My personal feeling is we can stand 30 seconds exposed to outside air at the cold airports, and it's not worth hundreds of millions to avoid this. I see no reason even the roll-up ramps can't be enclosed and given some heating or (with more difficulty) air conditioning.

Electric cars do get drained quickly by air conditioning. In some climates, other fuels might make sense. In addition, the tarmac can come equipped with inductive charging pads or other electric docking stations. Cars could go to these not just for a charge, but for any extended period of waiting to offer all climate control needed. It's also possible to cool cars by just having them stop every so often for a refill of ice water in a tank that does the cooling.

People with claustrophobia might not like travel in small cars. Bigger cars can be arranged for them. If they are going to handle the plane they can handle this.

In the full robocar world

Many of these steps change a lot in a world where robocars are on the ordinary streets, which I predict happens less than a decade after an airport like this can be built. The system adapts pretty well to the change, however. In the full robocar world there is a much reduced system on the non-secure side. Almost all the travel on the non-secure side would move to the owned or rented robocars of the passengers. There would be no more airport parking lots (freeing up tons of space) and robotaxis would handle all trips for visitors (most of whom would also not bother with a dedicated rental car, freeing up even more space.)

As I've noted, deliverbots would bring luggage to the airport in advance and take it to a passenger's home or hotel when it arrives, possibly beating the passenger.

Logistics and Cost

Most of the cost of this design of airport, aside from land, is the cars. In addition to being the transportation, the cars are also the waiting areas, the moving walkways and lounges. These cars should have a comfortable seat with a screen and a fold out table. There should be plugs for charging and wireless internet (ideally not at airport prices.)

The cars will come in a variety of sizes, based on analysis of the typical sizes of groups flying together. There would be many cars that are just for one person, and family cars that can hold 5 or 6. Of course, the airport would overprovision the 2 and 3 person cars and sometimes a solo flyer would get one. A proportion of the cars would be wheelchair ready, and some would be wheelchair dedicated.

Because the infrastructure of the airport is now made up of a fleet of cars, it can be expanded and shrunk to meet demand. Airports whose loads are seasonal might import and export boatloads or trainloads of cars to other airports with different loads.

In the full robocar world, only the cars for travel in the secured area would need to be operated by the airport. All the other transport would rely on the city's own robocars and taxis, which should have a great ability to respond to changes in demand.

At the periods of highest demand, the airport might switch to a mode where more passengers share cars if they are on the same flight. This introduces some delays for passengers as they won't get a personal car to the plane immediately, but will need to wait a few minutes for a car to fill up. Airports taking this approach would have a fleet that's mostly multi-person cars. The cars can be made almost as long as desired, as long as they are narrow enough to go through any tunnels, and can dock at their front or back rather than on the side.

A large airport with 30 million passengers/year (like San Francisco) is estimated to have a "peak hour" with about 10,500 passengers. While the average car load would probably be in the range of 1.5 normally, at the peak design hour the airport would probably switch to doubling up and reach an average of 2. If each passenger uses a car for about 30 minutes this implies a fleet of about 3,000 passenger cars. These numbers are initial guesses and I would need to learn more about airport flow models to do a real design.

For the first airport to do this, the cars would be more expensive, perhaps $50,000, though at these fleet sizes you start getting to the point where you can get real economies of scale and bring the cost down. The first airport would probably try not to have lots of different sized vehicles, and in fact would mostly try to build vehicles based on traditional road cars, fueled on gasoline and modified to be robocars. Ideally there would be a reasonably priced PHEV (plug-in hybrid) vehicle which could be modified, since a PHEV can switch to electric mode in order to go into indoor locations like security.

Initial service would not feature motorized doors, roll-in boarding for wheelchairs or any of the features desired long-term. It would look like a car or minivan with no steering wheel or driver dashboard and some extra sensors. All-weather docking becomes more complex here, and getting in and out of ordinary cars takes longer than walk-in pods.

However, with these numbers even a mega-city airport is built with $150M of vehicles. There would also be an annual vehicle replacement budget of around $25M as vehicles wear out. The rate of depreciation would depend on their duty cycle.

Other work

I'm not the only one to imagine an airport that uses automatic vehicles rather than terminals to get people to planes. PRT vendors have often pitched PRT for the outside-security needs of airports, and the first PRT deployment was a parking transport service for Heathrow Terminal 5's Business carpark.

You can read Martin Lowson's more full Heathrow analysis and this paper by Peter Muller contains many useful ideas that predate mine, though it still uses the concourse approach. He has another article with more details.

There was also a study of heavily using PRT around Sea-Tac airport.

Comments on this essay can be left at this blog post.