The idea of robot-like vehicles has been around for quite some time. We have seen them in movies and read about them in science fiction books. But how close are we really? With financial backing from the government and private industry, the engineering field has made tremendous progress over the last few years. However, the impact of autonomous vehicles in society brings up several issues that could hinder the industry’s progress.
The speed with which technology has revolutionized our lives in the last 20 years is truly amazing. Consider the fact that e-mail and the Internet have all but replaced trips to the library and the post office. Technology has even given us the option not to rush home to watch our favorite television programs; today, TiVo can take care of most of our entertainment needs. If you own a new car, you might have also experienced the seat that remembers your favorite settings and adjusts itself to fit you comfortably. As my mom would ask, “What’s next, a car that drives itself?” Actually, Mom, yes!
The idea of cars that can drive themselves is not a new one. The first exhibition was given in 1939 during the World;s Fair. At that time, General Motors set up a demonstration that had cars follow each other using magnetic floor strips . By today’s standards, it was not one of the most sophisticated presentations; however, it was a start.
Fast forward 65 years to the 2005 Defense Advance Research Projects Agency (DARPA) Grand Challenge, a 132 mile “unmanned” race. Only five cars crossed the finish line, the fastest car completing the race in 6 hours and 53 minutes with an average speed of 19.1 mph . Behind the wheel of the winning car were thousands of lines of programming code. The vehicle was driven by sophisticated software packages, mounted on the dashboard of the car, that made the decisions that you and I have to make every time we get into a car (where to turn, when to stop, etc.).
Let’s do it for our country
The Defense Advance Research Projects Agency (DARPA) Grand Challenge challenged various teams of students, engineers, designers, programmers and private sector partners to construct a vehicle that would cross the finish line with no human intervention. Imagine the potential benefits of cars that drive themselves: – More productive morning commutes – Fewer accidents – Better visibility during bad weather – All or most terrain navigation If you can’t imagine it yet, the United States government already has. In fact, the goal of DARPA is the development of unmanned vehicles to be used in the military. The Challenge and the resulting transformation of ordinary cars into robot-like vehicles capable of driving themselves is as much the result of ingenuity and technology as it is the result of a congressional mandate to make one-third of military land vehicles autonomous by 2015 .
The government has two major objectives for their autonomous car program. The first goal is to have follower convoys in military war zones. In addition to the efforts by DARPA, Robotics Technology Integration teams from the US Army along with the Development Engineer Center have been working on the Robotic Follower Advance Technology Demonstrator project . This Unmanned Ground Vehicle (UVGC) program would have the ability to transfer information from a manned vehicle to an unmanned vehicle that follows closely behind. The download of information from the first car would provide the robot vehicle a path that avoids areas that could impede or confuse its navigation system . Soldiers would continue to ride in follower vehicles but they would have time to sleep, eat or keep watch on the area. Steering the vehicle on the road and towards its intended destination would no longer be the responsibility of the crew .
The second goal of the UVGC program is the development of a completely self-sufficient vehicle. The uses for these advanced robot-like vehicles would be numerous when you consider war areas that are often dangerous to soldiers. The Army hopes to create all-purpose combat vehicles that would survey battlefields, sniff for land mines, or run supply missions without risking the lives of soldiers (Noguchi). Over time the program also hopes to build computer systems that have the ability to “learn” from mistakes and update their own code in order to avoid making the same errors twice .
There are evident uses of robot vehicles in the military, but this technology is relevant to our daily lives as well. In fact, while the DARPA Challenge has as its foundation in military goals, the technology can also be used to design a car that can one day navigate through our complex freeway systems. While government testing and goals revolve around the ability of a vehicle to navigate rough terrain and remember programmed paths, the goals of many of the participants of the DARPA Challenge were somewhat different. The real challenge was to give an autonomous vehicle the ability to control speed, direction and build a decision-making system that would allow it to mimic human decision-making.
Stanford University, the winner of the 2005 challenge, very quickly decided that a Global Positioning System (GPS) unit would not be able to handle the complex road obstacles that the race posed. GPS as a navigating system works well on constant flat terrain but has a difficult time interpreting quick, sharp turns and unexpected debris on the road. As a matter of fact, the Stanford team learned quite a bit from studying the 2004 race (a race it did not participate in); no vehicle had crossed the finish line. So the decision was made to add a more sophisticated and better suited navigation system to mimic the type of driving that people would do every day. And that is how Stanley, a modified Volkswagen Touareg, was born. The SUV looks very similar to regular vehicles, but notice the equipment that is mounted on the car. This equipment is vital to the technology Stanley used to win the Challenge: LIDAR, Light Detection and Ranging .
LIDAR is a versatile sensing tool. It works very much like radar except that instead of using broad radio waves, it uses narrow beams of light. LIDAR depends on the ability of its systems to calculate the speed of light when an optical impulse or light beam has been generated, sent and received. Light in the form of a laser is sent out ahead of the object. The system uses this time measurement to calculate the distance of various objects it encounters along its path by using the following formula (see Fig. 1) :
Once a measurement has been obtained, the system feeds this numerical information into a computer that can subsequently build a three dimensional map of the terrain ahead of the vehicle .
Would LIDAR, using radars, react any differently when tested during the DARPA challenge? Stanford felt that the answer was yes. This was one of the reasons Stanford chose a LIDAR unit over GPS as its navigating system; laser measurements tended to be more accurate and less affected by outside factors. For example, laser beams are not affected by time of day, angle of incidence or background noise. Additionally, because laser beams travel at the speed of light, they are unaffected by radiation, temperature variations, and pressure changes .
The effectiveness of LIDAR has also been the fuel behind Siemens’ Adaptive cruise control trials. This manufacturing industry player believes that the low cost and effectiveness of LIDAR sensors can provide accurate car-to-car distances. This will enable the system to intelligently modulate speeds based on traffic conditions and will allow the company to continue exploring similar driver-assistance functions.
Within our lifetime?
Although the technology is available and progress is being made, how realistic is it to expect cars to drive themselves during our lifetime? The experts say it is hard to put a time frame on this development. Rather than in years, they are more comfortable referring to engineering cycles. Akhtar Jameel, president and CEO of DaimlerChrysler Research and Technology for North America, estimates “about ten car cycles before you see anything resembling autonomous driving” .
This delay between the experimental and practical phases gives us time to analyze the financial and social concerns that necessarily come along with such advances in technology. The idea of leaving safety-critical driving decisions to a computer has raised very valid concerns from critics. Kai-Uwe Balszuweit, head of the driver assistance department at BMW, addressed these concerns by stating that “. . . a driver will not be freed from his own responsibilities, as the system does not intervene when it comes to the actual driving of the vehicle . . .” . In a later interview, Siemens representatives agreed by saying that drivers would always have the ability to turn software off “if they felt threatened – rather than helped – by it” . Others say that there is no need to be concerned. Autonomous vehicles will be “safer than today’s cars . . . no more dozing off behind the wheel, no more drunken driving . . .” .
Safety concerns aside, will we be ready to deal with the social issues and complexities that robot cars will bring to our lives? There are various issues we need to begin considering, issues ranging from legal responsibility when an accident does take place to the taxpayer’s acceptance of his money being spent on roads and technology that will (for a good number of years) be used by a very limited number of people . In the meantime, we will continue to benefit from the advances of this technology. Today, limited radar systems are already helping snowplows in Minnesota navigate stormy and visually challenging roads . Software is being tested that would stop short of making driving decisions for you but that would warn you if your vehicle was getting too close to the cars in front or if you were inadvertently crossing over dividing lines. Siemens’ driver-assistance system does this by using strategically-placed radars that monitor traffic patterns around your car .
And, the 12th Annual World Congress on Intelligent Transport systems held in November 2005, showcased two of Toyota’s newest and soon to be released gadgets: the Intelligent Parking Assist, a system that uses an ultra-wide-angle miniature camera to guide the car in self-parallel parking, and the Pre-crash safety system that uses radars and video cameras mounted on the car to detect imminent crashes and reduce collision impact .
Although a completely autonomous car might seem like it is still more than a few years away, remember how quickly we went from VHS to DVD or remember how fast stores replaced tapes with CDs? So, the next time you hear someone say “look, no hands”, turn around. They might not be talking about a bicycle.