Automated Public Vehicles : A First Step Towards the Automated Highway


Michel Parent

Abstract : we develop a new mode of urban transportation based on two concepts : the concept of car sharing so that the total number of cars in a densely populated area is decreased and made less polluting, and the concept of the automated highway so that these public cars can be made available at any time in a large number of stations placed along a network of dedicated and automated roads. Such a transportation system could offer an alternative to the private automobile but also to public transport systems in places or at times when mass transit is not efficient. In particular, we think that these automated public vehicles could use the existing lanes which are now dedicated to public transport such as the bus lanes or the tram tracks. If we extend the concept of car sharing with the concept of car pooling, this new transportation system could even one day replace some of the mass transit systems with a much better level of service.


Everywhere, the private automobile is now perceived as the most desirable means of transportation for short to medium length trips (say from 500m to 300km!). It is reasonably inexpensive, it is rather safe, it boosts the ego, it offers a private place for many purposes, it can be used to carry or store large objects (or animals), etc. But mostly, a car is synonymous with freedom and this is what makes it so desirable.

However, the automobile alone is not efficient enough to move everyone as they wish. As soon as the density of movements becomes too large, congestion sets in and travel times increase dramatically. The solution of this problem of freedom to move in large cities lies in two directions : discourage the use of the automobile (through cost, congestion or legislation) and offer more efficient alternatives (mass transit, car pool, shared taxis, etc.).

Although mass transportation systems are constantly being improved (see for example the automated metros such as the VAL), more flexible means are also needed, in particular at places or times where the demand is too low to justify mass transport. Among these flexible transportation means, we can mention taxis, dial-a-ride mini-buses (or shared taxis) and PRT (personal rapid transit) which seem to be coming back.

However, all of these alternatives have not proved their economic feasibility and remain marginal. A new concept is now emerging which may offer the same convenience as the private automobile and at a low cost : it is the concept of a self-service-car (also called station car if its main purpose is to complement a mass transit system) : a single public car used several times during the day by different users. Besides, if the vehicles are well adapted to city use (small size, electric engines, controlled speed, etc.), the problems of automobiles in cities may be minimized. This concept has been developed in France with the PRAXITELE system [1, 2].

However, for these self-service cars to be really attractive as an alternative to the private automobile, they must be available all the time at closely spaced locations. This implies that the cars must be moved according to the demand. Two techniques can be used to move the waiting cars : use employees or use automation or a combination of the two. We believe that a large number of "stations" could be placed along a network of "automated roads" where the cars could move automatically.

Furthermore, these self-service cars could be made available outside the automated network if they are manually driven. This is the concept of "dual mode" developed by INRIA [6] which could allow for door to door service. With some help from employees which would drive the cars to and from the stations which are on the automated network, this system could cover a very large area. Later, when the network of automated roads is sufficiently large, it could also be opened to private cars, as long as they are certified (and probably maintained) by the operator of the network.


We all know the advantages of mass transportation, be it trains, metros, trams or busses. All these means are highly efficient in terms of number of people transported per unit of space or energy, as long as the demand is sufficient. If the demand decreases, the operation cost remains the same and the system loses money. This is why most mass transit systems stop at night and also sometimes during off-peak hours. This is also why mass transit is not efficient in suburb-to-suburb routes: the demand is not sufficient along any particular route because the customers would have to walk too far at either (or both) ends of the trip. Each mass transit system has a certain operating range in terms of passengers per hour which cannot be crossed because of economical constraints (lower bound) or technological constraints (upper bound).

On the other end, we have the private automobile which is very efficient in a different range of demand. As long as the demand is low, cars and roads are an efficient system for regular transportation of people or goods. However, when the peak demand exceeds a certain density, congestion sets in and even with the construction of expensive infrastructures such as urban freeways, the economics of road transport deteriorate rapidly. Furthermore, many undesirable effects are added such as air and noise pollution.

If we want an efficient alternative to the private automobile, we must provide a public transit system which offers the same level of service. This can only be achieved through a combination of mass transit for the high flows and flexible public transport for the times or places where mass transit is not appropriate. Existing flexible systems are : taxis, dial-a-ride services, self-service cars and PRT.

1- Taxis

Taxis have to face several problems. The most important is certainly the problem of the demand level. If the demand level is not sufficiently dense, it means that each taxi has to reach a much greater distance (bad for economics and for service time) or accept to make very few trips per day. In this latter case, this activity can only be a complement to other activities and cannot justify itself economically (this is seen in rural areas where the taxi driver can have another job).

The second problem is the demand allocation. If several taxis operate in the same region (which can be good for economics and for service), some form of allocation must be made. Until now this is either done by  roaming, or by queuing, or by dispatch. The main problem with these approaches is that they are probably not optimal in the sense that the taxis are either too numerous, or not enough, or not at the right place. All this means that the economics are not very good (taxis wait for customers) or the service is not very good (all taxis are busy or too far away). New computer-based systems using GPS localization and digital communications try to bring some form of better management but the major problem is the cost of the driver when the taxi is empty.

2- Dial-a-ride services

These services have been put into place more than 20 years ago to replace bus lines in low density zones or time periods. Instead of running the busses regardless of ridership, it was decided to run them only according to the demand. Furthermore, the line could be changed according to the demand. Actually, all the lines could be fictional and the customer would have the possibility to go from any place to any place (as with a taxi) but without any guarantee on the time taken and of course not in a private way.

A major problem is the complexity of the routing problem if one consider the more complex system of door to door (or even station to station) transport in a large environment. However, these problem can now be solved efficiently with current low cost computer technology. The main problem however, remains the cost of running such a system. If the demand is too low, it will not pay for the cost of immobilization of the vehicle and its driver. This explains why such systems are not commonplace and often reserved for reduced-mobility persons.

3- Self-service cars

The concept is based on a fleet of cars which are available to a set of subscribers in specific « stations » for use in restricted areas. At the end of the trip, the car is left in another station. The cars are available 24 hours a day and the fare is based on the time the vehicle is used and eventually the distance it travels.

These systems are in the planning stage for demonstration in several cities in Europe [4-6]. In America, a similar principle is also being tried with « station cars » which are aimed mostly at the end trip to or from a mass transit system. The technological advances concern the electric car which is favored by city officials who want to rid the city of polluting cars, smart card technology which allows simple access control and billing, positioning of the cars by GPs to keep track of their location and digital communication with mobiles for the management of the cars.

The main advantages of this system are the following :

- relatively low investment limited to the acquisition of cars, the installation of the parking lots and the management center,

- relatively low operation cost since the cars do not cost much when they are waiting,

- good quality of service as long as the stations are close to the potential demand and the system is properly dimensioned,

- very good comfort and privacy, close to that obtained with a private automobile, without the constraints of ownership,

- good image for the city and for the user if the cars are electric.

On the other hand, the system has some disadvantages :

- one needs to have a driver license to use the system (not accessible to youngsters and to disabled persons),

- does not work economically if the demand is too low or too pendular.

4- PRT

The concept of PRT is very old and the first experiments took place - namely in the USA - in the sixties. The ambition of the PRT is to offer a public service with the same convenience as the private car :

- large number of stations, even in low density zones, so that one does not have to walk far to get a car,

- good interconnection between lines so that the travel time is minimized,

- automatic vehicles so that anyone can use them and also get high throughput on the line,

- small vehicles allowing comfort, privacy, and low operation cost.

The constraint of automatic driving imposes the choice of a dedicated track, completely separated from traffic and pedestrians. This constraint brings the major difficulty of the concept : how to build these dedicated tracks. The obvious solution is to make them elevated but this is costly and brings lots of resistance from people living nearby. Underground solutions are also very costly and not as nice to the riders. As for tracks at ground level, this brings the problem of the « barrier effect » which cannot be accepted all over a city.

Besides the cost of the tracks, the PRTs are also faced with the cost of the vehicles and of the control systems. Equivalent systems such as the VAL in France (an automated light rail) show that these costs cannot be underestimated and this will bring the cost of each car way above the cost of a standard automobile.

5- Synthesis

Here is a synthesis of the attributes of various individual public transportation means in the city :

Taxis Dial-a-Ride Self-service PRT
Investment Low Low Moderate High
Operations cost High High Moderate Moderate
Performance* Low to High Low to Moderate High High
Comfort High Moderate High High
Accessibility** High Moderate Moderate to High Moderate
Fare High Low to Moderate Moderate Moderate
Disabled persons Not welcome Special services Not possible Yes

* waiting time + travel time

** distance to walk at each end of trip


Our own concept is to combine self-service cars with PRT technology in order to offer automated travel where the flows can be high at certain periods and manual travel locally next to the automated network.

With the development of assisted and automated driving techniques, it is now feasible to consider the possibility to move cars automatically on paved roads between stations where they can stop and move to the normal road network. If the speed is limited (to say 15 km/h), the road does not even have to be protected. This means that stations can be on level ground and hence be very inexpensive compared to aboveground or underground stations needed with PRT. For higher speeds (for example between stations), it will probably be necessary in densely populated areas to have a protected network which could for example double an existing highway network.

The development of these automated vehicles could come from three different directions. The main and obvious one is the development of driving assistance and automation in regular cars. very rapid progress has been made in the last five years and there are many operational prototype cars now available which can drive autonomously on paved roads with a minimum of infrastructure. All these developments are now labeled 'Automated Highway Systems" (ASH) but car manufacturers, highway builders and governments are wondering how to introduce such a technology on a wide basis. Recently, researchers from Berkeley have proposed that a public transit authority could be the first user of such a technology in the most congested cities.

The second direction may come from the transit industry. In Europe, the concept of "intermediate transit" is under development with mass transit systems which could fill the gap between busses and light rail. The idea is to have a kind of small tram (possibly electric or hybrid) but on wheels. These vehicles would be (at least on some part of the trip) guided on a dedicated track. In order to minimize the infrastructure cost, the guiding technology under development would be non-contact. This is the case of the CIVIS project under development by Renault and Matra in France.

The third approach is to develop a new type of vehicle dedicated to this concept. This is what has been done at INRIA with the CyCab vehicle . This vehicle represents the first prototype of a new generation of vehicles completely under computer control but with a possibility of manual driving through the computer (drive by wire). The CyCab has been designed specifically as a public vehicle with manufacturing and maintenance costs in mind. This first generation of public vehicle has been aimed specifically at small distance and low speed trips that can be found for example in pedestrian areas such as those in historic cities. Other types of vehicles (larger and faster) can be derived directly from this model, the controls being the same.

The vehicle can now be driven either manually, or teleoperated (7), or driven in platoons with a single driver. We are now developing the techniques for autonomous driving on dedicated tracks using magnetic or optical (or both) markers on the road.


The challenge is now to define where these computer controlled vehicles could run automatically. The main difficulty is the interactions with other vehicles. The interaction could be minimized if the private cars are banned or accepted at very low speed and without any priority as this is already the case in the historic center of many European cities. The automated vehicles could also be restricted to operate in the automated mode only on locations where these interactions can be avoided such as special lanes parallel to main throughways. In the case where the automated cars must cross regular traffic, there can be an overpass or underpass which would be of very small size since the public vehicles will be rather small and non polluting. Another alternative would be to have a level crossing protected by street lights associated with a video monitoring of the intersection.

We can see that we have a trade-off between restricting the use of regular cars and making special infrastructure dedicated to the automatic cars. However, this problem has already been studied in many places where a priority to mass transit has been decided. In many cities (in Europe in particular), the circulation of private cars has been severely restricted to offer priority to light rail or busses. Similarly, dedicated lanes for busses are more and more set apart in order to give an advantage to mass transit over the private car.

This strategy of giving priority to mass transit can be justified on the grounds that space is used more efficiently and nuisances are minimized. A bus on a dedicated lane requires about 300 m of bus lane at peak time (one bus every minute at 5m/s) for 50 passengers or about 18 square meters for 1000 seconds per passenger for a 5 km trip. The same passenger using a car would need about 20 meters of lane for the same amount of time plus two (or more) parking spaces of about 30 square meters (including access ramps) for a total of 24 hours. The space-time factor would then jump from 18,000 to 60,000 just for the trip plus 43,200 for parking (parking space for a bus is negligible considering the total number of passengers it carries). Car pooling aims at divising this enormous space-time factor by 3 or 4 and hence making it close to bus on a dedicated lane. Of course, a light rail would have a much smaller factor as long as the frequency remains the same and the capacity of the train is increased largely.

Now, during off-peak times, the busses must seriously decrease their frequency in order to have a reasonable cost per passenger. They may even stop altogether at certain times of the day or the night. This is where the automated public cars could take over. Since the lane already exists, the cost of the equipment to make it an automated lane would not be very high if there are not too many crossings which must be transformed by under or overpasses or protected by special lights. This could then provide at fairly low cost a high quality service which would run 24 hours a day on demand without requesting more space.

Furthermore, if the vehicles are designed to be shared at peak time, they could advantageously replace the busses. Indeed, if we have an average of 4 passengers per car and cars are platooned in packs of 20 (as it is envisioned in the automated highway systems) with interdistances of 300 meters, this would mean a capacity of 4,800 passengers per hour instead of 3,000 with the busses, if the speed is the same. Besides, the average speed could be increased if the cars stop only when and where needed (they would be derived from the main flow). Advocates of the automated highway even consider as possible, maximum flows of 6,000 cars per hour at speeds of 100 km/h but this does not seem realistic in an urban environment with many stops.


[1]- Laramée Michel, Benéjam-François Evelyne. « PRAXITELE. Pour un Transport Public PRAtique Individuel ELEctrique ». Textes des Communications de la Journée CFE. La Rochelle, Novembre 1993.

[2]- Parent Michel, Dumontet François, Texier Pierre-Yves and Leurent Fabien. « Design and Implementation of a Public Transportation System Based on Self-Service Electric Cars ». IFAC/IFORS Congress. Tianjin, China. Aug. 1994.

[3]- Augello Daniel, Benéjam Evelyne, Nerrière Jean-Pierre and Parent Michel. « Complementarity between Public Transport and a Car Sharing Service ». First World Congress on Applications of Transport Telematics & Intelligent Vehicle-Highway Systems. Paris, France. Nov. 1994.

[4]- Allal Chafik, Dumontet François, Parent Michel. « Design Tools for Public Cars Transportation Systems ». Fourth International Conference on Applications of Advanced Technologies in Transportation Engineering. Capri, Italy. June 1995.

[5]- Daviet Pascal, Parent Michel. « Platooning for Small Public Urban Vehicles ». Preprints of the Fourth International Symposium on Experimental Robotics, ISER'95 Stanford, California, June 30-July 2, 1995.

[6]- Parent Michel, Fauconnier Sylvain. « Design of an Electric Vehicle Specific for Urban Transport ». Congrès EVT'95. Paris, Nov. 1995.

[7]-Bensoussan Stéphane, Parent Michel. "Computer-aided teleoperation of an urban vehicle". ICAR'97. Monterey, USA. July 1997.

Michel Parent can be reached at Informatics for Advanced Road Transports, INRIA, BP 105, F-78153, Le Chesnay Cedex, France. Tel: 33 1 39 63 55 93; fax: 33 1 39 63 54 91.

home2.gif (1492 bytes)

Last modified: March 23, 2004