A Smarter Car
IBM wants to improve communication between cars, roads, and drivers.
Vehicles are getting smarter all the time, thanks to a combination of sensor and wireless communications technologies. Car manufacturers say that tomorrow’s drivers will be assisted by a wealth of safety information generated by vehicles that can talk to not only each other but to the roadway itself. But with so much data often comes information overload. And that’s why computing giant IBM has launched a project to help the driver get the right information at the right time.
IBM calls the research initiative collaborative driving, and the company says it’s designed to prevent accidents and reduce traffic congestion. The work will be spearheaded by the IBM lab in Haifa, Israel. “More than a million people die on the roads every year around the world, and people waste a lot of time and money sitting in traffic jams,” says IBM researcher Oleg Goldshmidt. “You would like to help with both problems in any way possible.”
For IBM, that doesn’t necessarily mean creating new safety gadgets. Instead, it means bringing the principles of computer science to bear on the different systems already making their way into vehicles. “We start with the assumption that in a relatively short time, vehicles and roads will be equipped with sensor and communications technologies,” Goldshmidt says. He notes that there are already projects in the United States, Europe, and Japan that are testing roadway sensors that can wirelessly relay information about accidents, traffic congestion, and weather to an onboard interface in the car. “Let’s say your car is now more fully aware of what’s going on around it. The question is what information is most important to the driver, and how that information gets relayed..”
Goldshmidt says that through a combination of computer modeling and driving simulations, the company can better determine how all the data generated by today’s high-tech cars and roadways can be gathered and organized, then processed and prioritized in a way that’s most helpful to the driver. He uses the example of two smart cars approaching an intersection: “Maybe there’s an algorithm that figures out the safest, most efficient procedure to let those vehicles pass through without danger or conflict. We’re trying to find that algorithm.”
IBM isn’t the only company searching for that algorithm. Jim Misener, transportation-safety-research program leader at Partners for Advanced Transit and Highways, administered by the University of California, Berkeley, says that there is an entire field devoted to this kind of research: it’s called arbitration. He points to the U.S. Department of Transportation’s Integrated Vehicle-Based Safety Systems. The IVBSS system, Misener notes, “takes data from autonomous sensors” and then tries to prioritize, or arbitrate, among the various pieces of information.
But figuring out how to prioritize all this road data is no small task, according to Mike Gardner, director of Intelligent Systems Research at Motorola. His group has been dealing with similar issues as it tries to develop handheld devices that can interact with smart-car technologies. “A smart vehicle has to collect all this raw sensor data, fuse it, and then analyze it with something like pattern recognition,” says Gardner. “Then it has to decide, ‘Is that a person in front of the car, or is that just fog?’”
There is also the issue of competing emergencies. Let’s say a car is equipped with both a forward-collision warning system and a lane-departure warning system. What would happen when a driver changes lanes to avoid a stalled car? “In that situation, you wouldn’t want a lane-departure warning going off while you’re swerving into the other lane,” says Tim Brown, team leader of cognitive systems at the National Advanced Driving Simulator, at the University of Iowa. “So trying to figure out communication between warning systems such that certain warnings get suppressed under certain circumstances is critical to providing the driver with the information he needs to respond appropriately in a collision event.”
Brown thinks that integrating different warning systems is critical. But he points out that “military and commercial pilots are used to all these alarms going off, and because they have hours of training, they are able to respond. You can’t expect the average driver off the street to respond in the same way as a pilot.”
IBM’s Goldshmidt agrees. “We don’t want drivers to have to undergo extensive specialized training to drive a car. We think, in certain situations, we can help a driver with information and feedback.”
“But,” he adds, “there are still many things–such as analyzing complex situations with many inputs and factors–which people still do better than any kind of machine. We have no intention of taking that away from you as a driver.”