This article reviews the research and development of automated connected vehicles that aim to reduce road accidents, money, fuel, and conserve environment. Major automotive companies have added automated functions to their vehicles, and various driver assistance systems—adaptive cruise control, video-based lane analysis, and steering and braking assistance—are currently available on high-end models. Automated systems can assess some traffic situations faster than humans can. As a result, automated driving is expected to significantly reduce accidents and traffic fatalities, improve traffic flow and highway capacity, achieve better fuel efficiency, and reduce emissions. However, on the way towards fully automated driving, many challenges need to be addressed. There are technology issues, including reliability, and non-technical issues of cost, regulation, and legislation. In order to accelerate the development of fully automated connected vehicles, there is a need for a cooperative approach. A practical evolutionary roadmap can be developed by an interdisciplinary panel of experts representing major car companies, government agencies, research centers, and academia.


General Motors’ Futurama exhibit in the 1939 World's Fair included a vision of what transportation might look like 20 years in the future. The forecast for 1959 included self-guided cars and automated highways.

GM's forecast was a bit hasty, and its full vision has yet to be realized. But developments since then, especially in the past four decades, have brought us closer. Major automotive companies have added automated functions to their vehicles, and various driver assistance systems—adaptive cruise control, video-based lane analysis, steering and braking assistance—are currently available on high-end models.

Automated systems can assess some traffic situations faster than humans can. As a result, automated driving is expected to significantly reduce accidents and traffic fatalities, improve traffic flow and highway capacity, achieve better fuel efficiency, and reduce emissions.

Cars will be able to drive safely while close together, with far less need for wasteful acceleration and braking. If people can be entirely relieved of driving and navigation tasks, they will be able to talk on the phone or text without causing accidents.


Human error is the most costly influence on the road. The National Highway Traffic Safety Administration reported 32,788 traffic fatalities in the U.S. for 2010 and estimated that 93 percent of them are attributable to human error.

An advocacy group, the Association for Unmanned Vehicles International, published an estimate in May that traffic congestion alone costs the U.S. economy more than $85 billion a year.

Still, the public needs to be convinced that the technology is useful, desirable, and safe. There are also cultural, social, legal, business, and other issues that need to be resolved before society accepts fully autonomous cars. But there are already signs of change. Nevada has become the first state to officially license autonomous vehicles. California meanwhile has confirmed that the state does not prohibit or specifically regulate the operation of autonomous vehicles and that rule-making is required before 2015.


It isn’t Futurama, but automated driving is here. Autonomous systems help with steering and with path planning, including acceleration and braking. High-precision navigation, using GPS, other sensors, and databases determine the route to a destination. They all make driving easier and support the driver, who remains in control.

The level of automation can vary. The human driver may be required to monitor the system and be prepared to take over complete control of the vehicle at any time. In highly automated and fully automated modes, the driver becomes the operator of the vehicle, and does not need to monitor the system as long as it is active. In case of a takeover request, the driver has a certain time to react before the system returns to the minimum risk condition (standstill) by itself.

Accomplishments in Autonomous Vehicle Technologies

In the last four decades several attempts have been made in the United States, Europe, and Japan to develop car automation systems.

A simple self-driving vehicle, built by Japan's Tsukuba Mechanical Engineering Laboratory in 1977, could track white street markers for up to 50 miles and reach a speed of 20 miles per hour. A vision-guided Mercedes-Benz automated van, built in the 1980s by Ernst Dickmanns and his team in Munich, could reach a maximum speed of 60 miles per hour on streets without traffic.

The first automated research vehicles that were able to move without human intervention in heavy traffic for long distances were developed through the Pan-European Eureka Prometheus Project. The twin autonomous vehicles, VaMP and VITA-2, which were built in the 1990s, could reach speeds exceeding 110 miles per hour on the German Autobahn. They used computer vision to recognize very different kinds of objects and automatically avoided them.

The National Automated Highway System Consortium successfully demonstrated in 1997 the use of a fleet of more than 20 fully automated vehicles on a seven-mile stretch of highway in San Diego, Calif., before a gathering of transportation professionals and public officials. Because of budget constraints, the U.S. Department of Transportation canceled the program in 1998.

In 2000 Japan's Advanced Cruise-Assist Highway System Research Association demonstrated the effectiveness of magnetic sensors in driver assistance systems on the road using a group of 38 cars, buses, and trucks.

Between 2000 and 2002 a consortium of 12 car manufacturers, suppliers, and research institutes were funded by the European Commission to work on the Carsense project. Stock cars were used as a platform and were augmented with wide range of sensors to enable fully automated driving. Prototypes were successfully tested at slow speeds in dense traffic environments, and around urban areas with traffic jams, tight curves, and crossings, in the presence of pedestrians.

Google's driverless car fleet and various autonomous vehicle competitions—such as the DARPA Grand Challenge in 2005 and the Urban Challenge in 2007, with successful entries from Stanford University and Carnegie Mellon University, and VisLab Intercontinental Autonomous Challenge in 2010—have helped attract creative minds to the development of autonomous car concepts. As of 2012, the Google fleet included five Toyota Priuses and one Audi TT, which continue to navigate the highways of California and have logged over 200,000 miles (with occasional human intervention).

Researchers at MIT recently developed an “intelligent co-pilot” system that enables the car to share control with the driver. The system monitors the driver's performance and makes behind-the-scenes adjustments to avoid collision with obstacles and to keep the vehicle within a safe region of the environment. If the driver is distracted, and is about to run into a barrier, the system takes over, and steers the car back into a safe zone. The system is considerably lighter in sensors and computational requirements, and therefore is less expensive, than current fully autonomous vehicles, like Google's driverless car.

In summer 2012 the U.S. Department of Transportation started an extensive pilot test for vehicle safety communication in Michigan. It equipped 3,000 vehicles. First results are expected next year. A target date may be announced next year for the deployment of this technology to the general public.

The goal of automation is to create intelligent vehicles that can manage different situations and to connect all traffic participants with each other in some way. Connected vehicles have the potential of improving situational awareness, safety, mobility, and the environment. Vehicles would be able to inform drivers of roadway hazards and dangerous situations they cannot see. A vehicle driving over a patch of black ice, for example, can warn other vehicles and also alert safety crews to treat the hazard.

In an intelligent vehicle, a large suite of sensors monitors the state of the vehicle and its surrounding environment. A positioning system takes data from the sensors and provides information about the location of the car on the planet, on the road, and in the lane.


A motion planning subsystem guides the car in its second-to-second movement so that it can abide by traffic rules and avoid collisions. Planning tasks include automated parking, and adaptive cruise control, using radar to control velocity and distance from the preceding vehicle.

A control system translates decisions into actions in a fraction of a second.

On the way towards fully automated driving, many challenges need to be addressed. There are technology issues, including reliability, and non-technical issues of cost, regulation, and legislation.

The relation between the cost and the reliability of the autonomous subsystems is a major challenge. Some of the technologies used in current research of automated vehicles are reliable but very expensive. Attempts are being made to use more economical substitutes, in a smart way, to achieve the same reliability. A ‘temporary auto pilot” system recently proposed by Volkswagen is a step in that direction.

Experimental automated vehicles use lasers, radar, cameras, and satellite-based positioning systems to determine the vehicle's location and immediate environment. Wireless communication between vehicles and with conventional base stations has become a reality.

The evolution to automated production vehicles requires turning the imaging systems into automotive-grade series applications. Unlike human eyes, most sensors have difficulty adapting to changing light conditions, shadows, and different background color. Moreover, there is a lack of technology that can put the perceived scenario into context and make decisions based on previous experience—skills that basically make a human a good driver.

The laser sensors in research vehicles use a mechanism for rotating the laser beam and for providing a 360-degree view of the surrounding environment. The mechanism is not likely to meet automotive industry requirements of being maintenance-free and unaffected by dust or weather. The challenges for the laser sensors are important, since in all automated research vehicles those sensors provide the precise information about the contour and distance of objects. To date, no alternative to laser sensing has been identified.

Sensors can provide accurate information regarding an object's distance, contour, and potentially, its material. However, more context information is needed to determine the actual driving situation, its severity, and the necessary action by the car.

Current automated research vehicles cannot navigate in construction zones, accident areas, or other unexpected situations as well as an experienced human driver can. Current automated driving systems are not adequate to make decisions under uncertainty. For example, the images generated by laser, radar, or camera of a plastic bag blown across the street and of a ball followed by a child might be similar. However, the action taken by the vehicle should be very different in the two cases, but that is not possible with present-day automation systems.

Today's traffic relies on the human capability to understand context in very different situations.

Roadmap for Automated Vehicles

In the last few years, the automotive industry has introduced systems that can alert a driver when the vehicle comes too close to lateral lane markers or when another vehicle is in the blind spot when the driver is changing lanes. While both systems currently provide an audible, visual, or haptic warning to the driver, an active intervention with the steering or braking system of the vehicle remains in the future—perhaps the near future.

From a technology standpoint, automated driving vehicles might be possible in the short term, but as we remarked, there are still many challenges on the way towards their broad deployment.

We have a general expectation of the introduction of automated vehicles onto the public highways. Our predictions assume that the deployment of automated vehicles will be evolutionary, without disruptive developments comparable to those involving mobile communication technologies. There are other forecasts, however, which are more optimistic and predict that fully automated vehicles may be deployed before the end of the present decade.

We expect that, in the next five years, research vehicles will continue to make intensive use of new autonomous driving technologies. Some vehicles will be able to drive in automated mode at close range and to intervene in hazardous situations.


Legal Aspects of Automated Cars

Although the proper development and deployment of fully automated vehicles can eliminate many human errors and significantly reduce the number of accidents, automation is not expected to eliminate all accidents or traffic violations.

An automated vehicle may travel above the posted speed limit because glare renders the sign illegible. An automated vehicle may park itself near a newly installed hydrant that was not included in the vehicle's map data.

Similar disconnection with the environment could lead an autonomous vehicle into an accident that causes injury and property damage.

In the presence of automated cars, identification of an “error” scenario is needed, along with legal and contractual liability standards. The standards are needed to identify the responsibility of the vehicle's manufacturer, service provider, operator, and other traffic participants, as well as the role of insurance.

Traffic rules, established with no automated vehicles in mind, may also need to be revised. In fact, the definition of “driver” would need to be reconsidered.

Definitions of “automated,” “autonomous,” and “driverless” would also be needed to establish legislation that regulates if, and under what conditions, automated vehicles may be operated on public roads and interact with other traffic participants.

Some efforts for definition, standardization, and legislation of automated vehicles have started. The State of Nevada established registration of autonomous vehicles in May. Other states, including California, Florida, Texas, and New York, are working on similar initiatives. The Society of Automotive Engineers has formed a committee for “On-Road Autonomous Vehicle Standards,” and the German Federal Highway Research Institute has proposed definitions for different levels of vehicle automation.

If an autonomous vehicle becomes involved in an accident that causes serious injury or a fatality, the public reaction will be hard to predict. The presumption of responsibility may rest on the manufacturer. Consequently, it is conceivable that corporations may be involved in litigation that not only puts a company's direct assets at stake, but can also reflect very harshly on its reputation. This weighs a corporation's motivation to manufacture innovative products with its need to avoid litigation.

Further evolution will occur in the adaptive cruise control system. Cars will be able to control the speed within the limitations set by the driver and by other information provided, including the road curvature and posted speed limits. The cars will also be able to avoid collisions with other vehicles and objects through automated braking. Also, augmented reality technology will be used in assisting drivers, perhaps by combining the navigation system with displays projected on the windshield to indicate the correct path when approaching an intersection.

For the mid-term, the next five to 20 years, three important technology components are expected to further improve driver assistance systems and usher in highly automated driving: better sensor and recognition technologies, reliable vehicle communication, and accurate map data to provide additional context regarding the driving situation. For example, if the vehicle travels on a freeway, cross-traffic and pedestrians are extremely unlikely. If a vehicle operates in the city or an exit ramp, the target speed for longitudinal control is significantly lower than on the freeway. Such information can help in achieving safe and smooth vehicle control without requiring the driver to be continuously in the control loop.

Although much of the information could be provided through today's in-vehicle navigation systems, the requirement to have constantly updated, if not real-time, information needs to be emphasized. Moreover, the network of the different control units of the future vehicle will be connected to base stations and cloud computing.

The long-term vision is for the human to simply enter a destination into the vehicle's navigation system and let the automated vehicle perform all driving tasks during the trip. The vehicle platform may evolve, with the steering wheel and the brake and accelerator pedals disappearing, and the interior modified to facilitate nondriving activities. The human can then pursue activities completely unrelated to the driving tasks, such as working, socializing, or simply resting.

Eventually fully automated driving might lead to traffic patterns very different from those we know today. As vehicles become self-driving, many new scenarios can be envisioned such as automated vehicles becoming a service, called as needed. They can be dispatched, with no one inside, to customers.

In the next decades, automated driving and inter-vehicle communication technologies will have a dramatic effect on the vehicles we use, on our mobility, the economy, and the quality of our lives.


Challenges for Fully Automated Driving

The current traffic system relies largely on the human ability to assess a driving situation and act accordingly. This leads to a certain level of comfort and trust with all traffic participants.

Although an automated vehicle can acquire more precise position and velocity information from its sensors than a driver in a human-controlled vehicle, it does not have the perceptive capabilities to comprehend the context of a situation, or to make decisions under uncertainty. Therefore, the interaction of humans with the automated vehicle is expected to result in a new set of challenges.

The human passenger in an automated car might not agree with the automated driving style, for example, and that disagreement could manifest itself in a dislike for the journey, motion sickness, or a feeling of discomfort or distrust. A passenger could simply ask a human driver to alter the driving style, but such a request would not be possible in an automated vehicle, unless some basic settings for the automated driving style can be adjusted by the user.

Asleep at the wheel? Someday, BMW says, that may be practical.

Asleep at the wheel? Someday, BMW says, that may be practical.

Human factors research in other domains where automation is widely used, as in aviation, can be helpful in addressing some of the challenges. Ultimately, consumers will be the ones to decide whether they trust or desire the automated capability enough to give up vehicle control.

Several challenges pertaining to traffic coordination need to be anticipated when automated and human-controlled vehicles interact with one another, for example, at a stop sign or in a traffic situation coordinated by law-enforcement personnel. Almost any traffic situation might be handled differently by an automated or a human-controlled vehicle, although the automated scenario might be, at times, equally safe or possibly safer. Law-enforcement officers must be able to communicate to an automated vehicle that it is supposed to stop and yield to cross traffic, even if the actual traffic signage says differently.

In-vehicle communications are expected to revolutionize road safety, and alleviate the challenges that arise from different forms of interactions between automated and human-controlled vehicles. Consequently, the U.S. National Highway Traffic Safety Administration plans to make a decision in 2013 concerning a possible requirement that automotive manufacturers support these communications in new cars.

The creation of a safe, interoperable wireless communication channel connecting all ground vehicles, traffic signals, and mobile devices would be helpful in preventing collisions, and in maintaining homogenous traffic flow. Also, a communication channel to provide traffic coordination, or safety relevant information, from a central traffic control unit or law-enforcement personnel to automated vehicles would be desirable.

The current trend of convergence of mobile electronics, cloud computing, communication, information, social networking, artificial intelligence, and other leading-edge technologies will turn future cars into more than just a means of transportation. Connected intelligent vehicles will have the capability to access or generate information, and share it with passengers, public infrastructure, and machines.

They will become human-centered intelligent systems with expanding possibilities. They will serve as mobile offices, virtual assistants, entertainment centers, and mobile advertisement platforms.

New business models and new value chains will be created. Neural and other novel interfaces will be added which enable commands to be given with muscle impulses, eye movements, sound waves, and brain waves, thereby creating an almost symbiotic relationship between humans and vehicles.

To accelerate the development of fully automated connected vehicles, there is a need for a cooperative approach. A practical evolutionary roadmap can be developed by an interdisciplinary panel of experts representing major car companies, government agencies, research centers, and academia. The work of the panel can be facilitated by using an intelligent system that enables crowdsourcing to get ideas from a key segment of stakeholders—the public in general.

Note: Readers interested in pursuing the subject covered in this article will find more information at

The website, created as a companion to this Mechanical Engineering magazine feature, contains links to material on intelligent vehicles and connected mobility, and to some of the major research centers at academic institutions and automotive industry that are working on related technology.