Hamilton-based McMaster University has teamed up with IBM to study how using a single multi-core processor to connect multiple microprocessors in automobiles will improve operating efficiency and driver safety.
The plan is to integrate the vehicle sensors and microprocessors—which currently don’t share information—using a “super” microprocessor, originally developed for video games.
Currently there are anywhere between 40 and 100 processors in cars, usually cheaper units that have very specific purposes. “We are looking at replacing some of them with something a powerful multi-core processor and to take information from current processes and get an overview of functionality,” says Alan Wassyng, project leader and acting director of the new McMaster Centre for Software Certification and McMaster’s Software Quality Research Lab.
To date, Wassyng’s research has been focused on safety-critical software for nuclear energy and medical devices.
In the nuclear industry, safety critical applications have redundancy built-in. Basically several computers process the same information to determine how the data should be interpreted.
“We can’t normally do that in a car. But if you had these multi-core processors, you would have enough process capabilities,” says Wassyng. “We can be much more tolerant of hardware faults and detecting different aspects from different sensors.”
Competition is fierce, and many automotive manufacturers are looking into added safety functionality for future vehicle designs. Some higher end vehicles already have sensors that alert the driver of an obstacle in their blind spot and some have automated parking capabilities.
By using IBM’s multi-core heterogeneous processor design, Wassyng hopes to reduce the number of processors needed, while improving the communication lag between processors.
By increasing the computing power, vehicles will better integrate into transportation systems such as roadside and emergency services and traffic management.
The research team will use IBM’s Rational software—developed exclusively for the automotive industry—to design and manage software throughout the project.
In addition to the safety functionality, the researchers are looking at how to give drivers real-time visual information and alerts that suggest alternate routes when traffic is congested.
“There may be situations where you have a special engine for doing graphics processing, maps and heads up displays, and other processors that are adept at doing streaming data coming from a network in the automobile,” says Don Aldridge, general manager of research and life science at IBM Canada. “Some of those networks exist now. But if you start putting them all on the same chip, you’re communicating at chip speed [which is faster], one processor to the next.”