Kumar develops automated, collision avoidance software
Lauren Eichmann and Alexis Terrell, ECE Illinois
- Prof. P.R. Kumar and his students have been studying new dimensions of wireless networks, merging the fields of communication and computation with control.
- Kumar predicts that the future of wireless networks will resemble the cell phone industry in which large networks are created that can then be altered to work with the physicla world around it.
- A simplified version of this futuristic scenario is the vehicular test-bed constructed in Kumar's Information Technology Convergence Lab.
ECE Professor P R Kumar and his students have been studying new dimensions of wireless networks, merging the fields of communication and computation with control. Kumar anticipates the next phase of the information technology revolution to provide ad-hoc wireless networks, or those without an infrastructure similar to that of a cell phone tower, that can be customized to interact with the physical world and alter it.
Earlier this year Kumar served as a guest lecturer at the Royal Academy of Engineering in London where he explained his research and the technology being developed in his lab. He also participated in two BBC radio shows in conjunction with his lecture.
“We are entering an age where wireless allows entities to communicate and pass on information to each other and thus interact with each other,” said Kumar. “One can envision that the twenty-first century will be the era of building large, smart systems.” He explained that as we enter an era of greater awareness of resource limitations and environmental issues, we want to be much more efficient in our resource consumption . This will involve the use of smart energy grids, automated highways, or sensor networks in hospitals.
During the lecture, attended by nearly 100 delegates, Kumar explained that a wireless network may someday prevent individuals from crashing their vehicles, among other things. If enabled, such a “vehicular network” would provide a way for cars to communicate with one another for collision avoidance.
In future traffic systems, for instance, cars may be controlled automatically along smart roads, improving travel time and reducing accidents. Vehicles would move along travel-time optimized routes connected by sensors, controllers and actuators that communicate over a network of computational nodes to avoid collisions and delays. A simplified version of such a futuristic scenario is the vehicular test-bed constructed in Kumar’s Information Technology Convergence Lab at Illinois.
Miniature cars, with color-coded roofs patterned to differentiate among them, are controlled by radio transmitters connected to laptop computers. Two cameras located on the ceiling provide vision feedback while additional desktop computers process the images on a data server. The server then communicates instructions to the cars controlled by individual laptops connected through a wireless network.
The collision avoidance algorithm works by having each car try to clear an area in front of it so it may gain approval to move into that space. If it gets denied, the car then stops and thus avoids a collision. In the Convergence Lab, Kumar and his students have demonstrated successful tests with up to eight cars driving in city traffic type situations, parking automatically, and even adapting to system upgrades and restarts without interrupting traffic flow.
Automated traffic, however, is just one example of what these smart systems can do, Kumar said. Cars could potentially be replaced with airplanes or power systems. He noted the system is not yet foolproof though. Potential liability and insurance problems may be associated with the system, and other factors like pedestrians and animals would have to be considered for maximum efficiency. The overarching goal of the IT Convergence Lab, Kumar explained, is to make it easy to build complicated systems that use computers and networks to interact with the real world.
“The goal is ease of customization,” Kumar said. “The world is highly embedded with computers. If we can learn how to easily design networked control systems, we can set the stage for massive proliferation. The obstacle to it is complexity both of design and operation. We want to make interfacing networks with the world as easy as putting together Legos.”
“Take your electric socket. You don’t really care what’s behind it. It could be connected to a hydrogen plant or solar cell,” he said. “You only want it to produce electricity. We want this simplicity in everything. We want to hide the complexity of the infrastructure from the designer.”
In Kumar’s research, the underlying support systems, called “mechanisms,” become just as important as the application of those services. “We’re solving a family of problems,” Kumar said. “We’re not just building a car but studying how you could build a system of cars, and use that same methodology for other systems. It’s the difference between studying a language and studying linguistics.”
One example of a system being investigated in Kumar’s Lab is “intelligent intersections,” where stop signs and traffic lights are removed, but cars are still able to effectively communicate their locations with one another and regulate the continuous flow of automated traffic. He said this would be especially beneficial at night and in rural areas when traffic is minimal and drivers often waste a lot of time and fuel on needless traffic controls. Such intelligent intersections can also enhance safety and reduce the health care cost of accidents. “Most car accidents are caused by human error,” Kumar said. “If someone could just tap you on the shoulder two seconds before you make a mistake, you could avoid crashing. That’s what I want — to make our lives safer and easier.”
Another potential use of intelligent systems could be in automated buildings that integrate lighting, alarm systems, and air conditioning. The “sensor network” system would detect when someone is in a particular room, and direct flow of hot or cold air, and turn on the lights. If there is smoke, it would even alert the fire department. This could also be applied in the home. An alarm clock could be interfaced with a toaster, or a microwave oven. Such sensor networks, composed of wireless nodes that contain sensors of the environment, could not only potentially detect fluctuations in temperature and light, but also, in other applications, the speed of a car, moisture, magnetic fields, acoustics, and even how many steps a person walked.
Kumar and his students want to see this technology applied everywhere in commercial, industrial, and civil applications. Kumar said the Convergence Lab has been supported by contracts with Toyota, Rockwell Collins, the National Science Foundation, DARPA, the Army Research Office, the Office of Naval Research, and the Air Force, among others.
“The Federal Communications Commission (FCC) has already allocated a spectrum for vehicular networks,” said Kumar. “So things are happening fast. We are very much in this type of world already.”