Road Map Towards Full Driving Automation Michel Parent Ming Yang INRIA -- BP 105, F78153 Le Chesnay Cedex, France [email protected] [email protected] Abstract Automobile has become the dominant transport mode in the world in the past several decades. In order to meet a continuously growing demand for transport, one solution is to change the control approach for vehicle to full driving automation, which removes the driver from the control loop to improve efficiency and reduce accidents. Recent work shows that there are several realistic paths towards this deployment: driving assistance on passenger cars, commercial vehicles on dedicated infrastructures, and new forms of urban transport (car-sharing and cybercars). Cybercars have already been put into operation in Europe, and it seems that this approach could lead the way towards full automation on most urban, and later interurban infrastructures. China has great potentials to apply cybercars technologies because of the needs for high capacity and low environmental impact. Beijing Olympic 2008 and Shanghai World Expo 2010 could offer perfect large scale demonstrations. 1. Introduction Throughout the twentieth century, the automobile and its infrastructure were developed in such a way as to become the dominant transport mode for passengers as well as goods in most countries. However, this development has been mostly quantitative with more and more new roads (and in particular new freeways) and more and more vehicles. Although this quantitative development has not reached its full extent in many countries (with enormous potential for growth in Asia), in most industrialized countries a saturation point seems to have been reached in terms of infrastructures as well as number of vehicles. In order to meet a continuously growing demand for transport, the solutions for industrialized countries now lie in better management of the resources (infrastructures and vehicles), in better use of intermodality but also in new technologies for vehicle control. Indeed, the control techniques for vehicles have not changed basically in the last one hundred years with the driver having the total responsibility of his vehicle through mechanical impediments (steering wheel and pedals). These primitive controls lead to inefficiencies and accidents. The only way to improve efficiency (defined as maximum throughput per unit of space) while at the same time drastically reduce the number of accidents is to remove the driver from the control loop. Obviously, this approach is needed for throughput only when and where the transport capacity has been reached, but in order to reach large improvements in safety, it should be extended to the largest part of the road network. The major problem, which has been encountered in the study of automated highway

1

systems, is the deployment of this technology. Recent work shows that there are several realistic paths towards this deployment and that the technologies are arriving at a faster pace than expected ten years ago[Parent M.1993]. We can now identify three paths, which could lead to full driving automation in large parts of the road network: -

Driving assistance on passenger cars, Commercial vehicles on dedicated infrastructures New forms of urban transport (car-sharing and cybercars).

We will see that these three concurrent approaches use the same technologies and can lead to a large interoperable road network in the long-term future. 2. Driving Assistance on Passenger Cars Car manufacturers are very actively introducing drive-by-wire technologies in their vehicles for numerous reasons : simplification of mechanical design, lower cost, better performances, and new functionalities. Among the new functionalities, many concern taking control over the driver for various situations: longitudinal and/or lateral control on highway, emergency braking, lateral control in case of skidding, parking assistance, lateral and longitudinal control in Stop&Go situations. The objectives of these new functions are to improve comfort for the driver and also to improve safety. It has been shown that these functions can also improve, in some cases, the throughput[Mc Donald 2002]. Many of these functions imply the use of advanced sensors: gyrometers, accelerometers, radar, lasers, cameras, ultra-sounds. Already, putting together these techniques could allow for fully autonomous driving in particular environments, for example on freeways. The difficulty lies in the unpredictable behavior of other (human) drivers or in the handling of situations with unexpected obstacles. Therefore, it seems difficult to avoid leaving the final responsibility to a human, at least if we cannot secure the infrastructure (which lead to chicken-egg problem). Driverless cars do not seem for the near future with this approach alone. 3. Commercial Vehicles on Dedicated Infrastructures Commercial vehicles are also being equipped by manufacturers with drive-by-wire technologies and advanced sensors (using the same technologies as for the private automobiles). The objectives are however slightly different. The main objectives for these technologies are the lower operation cost for the user. This means better fuel efficiency, better safety and lower personal cost. These are the major drivers for

Figure 1. Chauffeur

Figure 2. Phileas

2

Figure 3. Crayon

developments such as platooning where a single driver can drive a train of two or three vehicles (which can be trucks or busses, see fig. 1). The infrastructure operators (freeways or city roads) have another objective for these technologies: it is to improve significantly the throughput and to separate different kinds of traffic for better comfort and safety of the cars and vulnerable road users. In the case of commercial vehicles which run on limited infrastructures such as in industrial plants or harbors and also for mass transit vehicles, it can be advantageous for the operator to have a fully automated system on a dedicated infrastructure. Usually, on such infrastructures, the risk of collision with other vehicles or unexpected obstacles is very much reduced and therefore, fully automated driving is possible. Several such systems are already in operation or will soon be. At the moment, these dedicated infrastructures are very limited in number of kilometers but they could grow significantly in cities (as an alternative to tram lines) and in the longer term as truck or bus lanes on freeways. In cities where dedicated lanes for automated high capacity vehicles are starting to appear (see the development of BRT or Bus Rapid Transit such as the Phileas(fig. 2) which will run soon in Eindhoven), the future may very well be with smaller automated vehicles such as the ParkShuttle which could offer similar capacity with much more flexibility at a similar cost. In this approach, driverless vehicles are already on the road in limited environments and low speeds. However, with the development of dedicated infrastructures, higher speeds can be attained very rapidly and then the deployment will be at the pace of the deployment of these infrastructures, which will be limited if they do not serve a large portion of transport needs. 4 Car-sharing and cybercars In cities, the ownership of a private vehicle is becoming more a constraint than a pleasure. With the arrival of new forms of car rental (in particular with low cost operators) available throughout entire cities, citizens are turning away from car ownership and become truly multi-modal. The most advanced forms of car rental are the station-car systems, first experimented in the French Praxitele project. In these systems, the subscriber can pick up a car at any station and leave it at another station, therefore using the vehicle only a minimal amount of time, often just to reach a mass transport. Such systems rely on advanced real time management and wireless communication. Car manufacturers such as Honda now offer specific city oriented vehicles (small, low noise and ultra low emissions), which integrate communication and management technologies for such car-sharing systems. Cities which have to face the degradation of the quality of life brought by traffic, might also be tempted to restrict access to all non zero pollution vehicles if an alternative with specific, city oriented vehicles is offered by car-sharing organizations(fig. 3). The future of such systems lies in the development of cybercars which are vehicles with fully automated driving capabilities in order to have a real door-to-door service (cars can be called at any location and can be left anywhere because of their autonomous driving capability). The first cybercars in operation are specific urban vehicles, which run at low speed in restricted environments such as private parks,

3

campuses or pedestrian zones. The next generation of cybercars (now under development such as with the AMICA project from Fiat) will be city cars with dual mode capabilities: they run automatically on a restricted environment and manually elsewhere, just like a regular car (but with advanced drivers aids). The operators of such fleets will certainly be very interested to have access to dedicated infrastructures such as the automated bus lanes (see above) to extend the automatic range and increase the speed in automatic mode. If these vehicles become very popular, it will be a simple matter to develop a new dedicated (and protected) infrastructure (very light and possibly elevated such as in Cardif), which could have a high throughput because of platooning techniques, to link the automated zones together. Such infrastructures, which would form a loose network, could be very efficient to link suburban zones, and at a later stage for interurban travel. With the availability of a large network for automated travel, private cars would be interested to use these infrastructure (under control of the overall management of the resource and possibly with a toll) for automated driving, using the capabilities of the driver aids. For the same reason, freight transporters would also be interested to use the network for automated goods delivery, in particular for city environments where large (and perhaps medium size) delivery trucks are or will be prohibited. We can see that this is the most promising approach since it can serve a large portion of transport needs. If the political will be present to offer an alternative to traditional cars, this approach could lead quickly to a large infrastructure dedicated to driverless cars (with mostly existing roads for low speed driving in limited access areas) and new infrastructures for high speed driving. 5. Cybercars Applications 5.1 Existing application in Europe Several transportation systems based on cybercars are already in operation and several more are now at the planning stage, such as Floriade 2002 by Yamaha, ParkShuttle by Frog in Schiphol airport and Rivium of Netherlands, Simserhof Ride and CyCab by Robosoft in France, ULTra by Bristal in Cardiff, and Serpentine capsules in Lausanne(fig. 4) [Parent M.03]. Although these first systems are small in scope, it is believed that this will allow for the demonstration of the full benefit of such systems on a larger scale. 5.2 Potential applications in China As most European cities, big cities in China face numerous challenges associated with the use of private vehicles, like road congestion, energy expenditure, noise and pollution, all of which degrade the quality of urban life. An alternative flexible public transportation is necessary to solve these above problems besides the railway[Yang M.03]. China has great potentials in cybercars application, and Beijing Olympic 2008 and Shanghai World Expo 2010 could offer large scale demonstrations. In order to meet the huge transit demand in these two large events, the municipalities

4

(a) Floriade 2002

(b) Schiphol airport

(c) ULTra in Cardiff

(d) Simserhof Ride (e) CyCab (f) Serpentine Figure 4. Some existing applications of cybercars plan to build several new metro lines. However, these metro systems might lose money, since the operation cost is high and there is no sufficient transit demand after these events. The key to this problem is that the existing transport modes lack the flexibility in capacity. However, it is possible to solve this problem with cybercars technologies, which is able to provide a 24-hour, flexible public traffic service: The flexibility exists in multiple operation modes, adjustable capacity and individualized route selection: -

-

During the peak hours, cybercars can operate in platooning mode to maximize the transportation capacity. During the normal time, cybercars will be operated individually to maximize the flexibility and minimize the waiting time; Since there is no mechanical link in platooning, the capacity can be easily adjusted by changing the number of vehicles to satisfy different transit needs. The route can be dynamically adjusted according to the request by passengers.

Beside that, cybercars are electrically driven light vehicles with low velocity, and able to offer a safe, comfort, and quiet public transportation with low impact on the environment. Cybercars can provide individual service with on-board information system and modern vehicle design , especially for the elders and handicapped. More importantly, the application itself offers a very attractive demonstration to exhibit the innovative transportation system. Cybercars can improve the attractiveness of the Olympic and the World Expo by their advanced automatic driving and modern appearance, and can be integrated with the sites with little infrastructure cost. Besides, cybercars can be applied not only in large events, but also in resorts, parks, downtown, campuses, CBD, industrial parks, areas around public transport stations, etc. New application cases are coming forth with the spread of cybercars.

5

6. Conclusion The technology for driverless vehicles is now available but its deployment is limited by the need to share roads with existing users. If we accept and extrapolate the developments, which are taking, place in the three different sectors described above, we could end up in fifty years with a road network, which will be split into two. -

One network will be reserved (mostly, there can be some possible access to certain manual vehicles) to fully automated vehicles (public or private, for passengers or for goods), which will allow on demand transport with some form of management to avoid congestion. This network will concern mostly the existing streets (possibly redesigned) in dense areas, which have decided to limit the use of traditional vehicles, and new light dedicated infrastructures where the automated vehicles could run safely at high speed and high throughput.

-

The other network will be the “manual” network where automated vehicles will go under the control of a driver (with more or less assistance) and older (manual) vehicles will also be allowed.

It could be possible that, in modern countries, more than 90% (in terms of passengers or tons transported) of all trips could be made in driverless mode in year 2050. Since we start with a percentage which is next to zero (but not null), the growth rate of this sector in terms of vehicles and infrastructure should be enormous in the next ten to twenty years. Acknowledgment The CyberCars project is funded by the European Commission Programme IST (Information Society Technologies) and the CyberMove project is funded by the EC Program EESD (Energy, Environment and Sustainable Development). References McDonald Mike, Henry Alain, Espié Stéphane, Parent Michel, Vaa Torgeir. (2002). “Stardust – A Research on Deployment of Urban Sustainable Transport Systems”. ITS World Congress. Chicago. Oct.2002. Parent Michel, Texier Pierre-Yves(1993). "A Public Transport System Based on Light Electric Cars". Fourth International Conference on Automated People Movers. Irving, USA. March 1993. Parent M. and Gallais G. (2002). “Intelligent Transportation in Cities with CTS”. ITS World Congress, Chicago, Oct. 2002. Parent M. and Gallais G. (2003). “CyberCars: Review of First Projects”. 9th International Conference on Automated People Movers, Singapore, Sept. 2003. Yang M. and Parent M. (2003). “Cybercars: An Alternative Public Transportation for City of Tomorrow”, Sustainable Multi-Modal Transportation for Chinese Cites, Shanghai, Oct. 2003.

6