Return to Automotive Electronics Home Page

Developing the Crash-Proof and Self-Driving Car

by Wes Strawn, Clemson Automotive Engineering Graduate Student

Vehicle safety has become more and more important over the history of the automobile. There is, however, an alternative to protecting drivers in the event of a crash—preventing the crash in the first place. The idea is simple: what if a car was smart enough to be able to correct mistakes the driver makes? This project is an attempt to take a look at the current advances in automotive electronics that are leading towards crash-proof vehicles.

Current Systems

Great advances in electronics over the past few years have led to more and more automation in cars. This section gives short summaries of some of the most cutting-edge systems designed to enhance the driver’s abilities in preventing an accident.

Automatic braking systems monitor conditions ahead of the vehicle and can automatically apply braking pressure in the event of an imminent crash. These systems use a mix of radar sensors, high-resolution image sensors, and laser sensors in order to accurately identify potential threats to the automobile. Currently, these systems can only mitigate the severity of an accident—but Volvo will release a system in 2010 that can apply full brake pressure, stopping the vehicle completely at low speeds.

Lane departure warning systems monitor the markings between lanes on a road and also keep track of turn signal activation. If the system detects that the vehicle is crossing the lane markings without any signal from the turn signals, a lane departure warning is issued. Some systems can then apply a braking torque to the front wheels, steering the car back into its lane.

Adaptive cruise control systems monitor the distance between your vehicle and a vehicle in front of you, and automatically adjust your vehicle speed in order to avoid collision with the vehicle in front. These systems can control both the throttle and brakes in a car.

Blind spot detection systems use sensors on the side of the vehicle to inform the driver of an obstruction in the blind spots located over the driver’s shoulders. These systems use radar or image sensors to alert the driver with an indicator in the side mirrors.

Electronic stability control improves the handling performance of the vehicle and prevents loss of control during the severe driving maneuvers often necessary during accident-avoidance scenarios. These systems can apply different yaw moments to the car by applying different brake forces to each wheel as necessary, controlling slides and over or understeer. Some stability control systems can also regulate the throttle, giving the system control over both the brakes and the engine.

Dedicated Short Range Communication is a communication protocol being developed to allow for vehicle to vehicle communications. Giving vehicles the ability to communicate with each other could vastly decrease the amount of vehicle-vehicle collisions. DSRC can also be used to notify drivers of emergency vehicles and to enhance other systems, like adaptive cruise control.

All of these systems, when considered as a whole, give the car itself control over acceleration, braking, steering, lane selection, and even an understanding of the traffic environment around them. It is only a matter of time before all of these systems reach the level where, in the event of an imminent collision, the driver is virtually eliminated from the equation and the car itself prevents an accident—making the vehicle virtually crash-proof (system failures notwithstanding). Unfortunately, current vehicle electronic systems are still very fragmented—it is very difficult to create one common network that all vehicle electronic systems can communicate and collaborate across simultaneously.

The Self-Driving Car

The ability to eliminate human error in the event of an imminent collision presents a natural extension of concept—the ability to eliminate the driver altogether, creating a driverless car.

The advantages of eliminating human error in driving situations is obvious—again, system failures notwithstanding, all problems associated with driving would immediately be eliminated. DUIs, reckless driving, speeding, and accidents would be drastically decreased and eventually eliminated as the system gained popularity. Considering that there are over 40,000 automotive-related fatalities in the US alone each year, the ability to eliminate even a fraction of fatalities would be massively beneficial. Traffic signs and regulations could be eliminated or computerized and optimized, minimizing traffic delays and shortening commute times.

There have been several recent efforts with the goal of creating an autonomous car. The first of these was the Euro EUREKA Prometheus Project. This program lasted from 1987 to 1995, and succeeded in creating two vehicles, both of which drove long distances in heavy traffic. The two vehicles drove at speeds of up to 130 km/h and were capable of autonomously passing other vehicles and controlled convoy driving. These two vehicles did require occasional human correction, but managed to reduce the human intervention rate to once every 9 km (about 95% autonomous driving).

In 1995, Carnegie Mellon University had a project dubbed “No Hands Across America”, in which a vehicle managed to steer itself (the throttle and brakes were still human-controlled) over 3000 miles. The Carnegie Mellon car achieved 18.2% autonomous driving.

From 1996 to 2001, the University of Parma headed the ARGO project, which was the first car designed to follow the painted marks on an unmodified highway. The vehicle only used two black-and-white video cameras to keep track of the lines on the highway, and achieved 94% automatic driving.

The US Government has funded a series of projects dubbed Demo, with the latest (Demo III) capable of controlling its own throttle, steering, and brake, and communicating with other vehicles to coordinate movement. The demo vehicles were designed with a military slant, and were thus focused more on off-road navigation.

The first Defense Advanced Research Projects Agency (DARPA) Grand Challenge was in 2002. The Grand Challenge began as a competition between international competitors with fully autonomous vehicles over off-road terrain, but by 2007 it had evolved into an on-road “Urban Challenge”, with the off-road course replaced by a 55-mile simulated urban driving environment.  Prizes were given for the most successful autonomous vehicles; the prizes for the 2007 challenge were 2 million, 1 million, and 500 thousand dollars for the top three teams. The contest was won by Tartan Racing, a joint-effort between Carnegie Mellon and GM.

An alternative route to developing a driverless car has been to develop automated transportation systems that traditional cars can be linked into (with modification). The first of these, called Dual-mode Transit, allows otherwise normal, human-controlled cars to connect into a public monorail system for long-distance trips, not unlike the monorail systems at some airports. These monorails would be centrally controlled, allowing the automobile drivers to relax for the course of the trip and then disconnect at their discretion, returning to normal driving.

The next, called Automated Highway Systems, use a special lane on traditional highways with embedded magnets and vehicle-to-vehicle communication to allow vehicles to “link in” to the magnetized highway and then communicate car-to-car to maintain safe traveling distances without driver input. These systems seem to have lost popularity, and little development has continued since the late 90’s.

Engineers in The Netherlands have developed a hybrid technology called free-ranging on grid (FROG) that allows specially designed electric vehicles to simultaneously operate autonomously and communicate with a centralized grid system. The cars locate themselves independently and select their own path from point to point, but communicate their location back to the central system, which avoids collisions and redirects traffic as necessary to ensure optimum traffic flow. The major limiting factor in this system is that it is currently only capable of controlling around 100 vehicles simultaneously.

For More Information
[1] DARPA Grand Challenge, Wikipedia.
[2] Intelligent Transportation System, Wikipedia.

[3] Intelligent Transportation Systems, RITA web site.
[4] DARPA Urban Challenge, DARPA Urban Challenge web site.
[5] The Future of Automotive Electronics, Automotive Engineering International, 2009.