Modern cars and trucks are powered by software. That’s true for gas-powered vehicles, EVs, and hybrids. In any current model, software will play an important role in driver safety, navigation, entertainment and operations. Software controls the luxury options that make you comfortable, the safety features that help protect you, and even the systems that help your car accelerate, steer, and stop.
A typical new car has more than 100 million lines of software code – and General Motors is committed to prioritizing quality in every line in every vehicle we make.
While in some ways that makes cars and trucks more like PCs and mobile phones, in our case the stakes are far higher. We don’t ask drivers to serve as software beta testers.
GM’s quality labs work tirelessly to detect and address potential issues in our software long before it reaches the customer. Our software team includes experts from the largest tech giants who helped develop some of the best-known digital products on earth. Engineers come to GM from the tech giants to take on some of the world’s biggest, most interesting technical challenges.
GM has increased its push to leverage the power of artificial intelligence, and that is particularly true when it comes to software. We’re using AI to run software tests around the clock.
Automated testing simulates millions of button pushes and screen taps, helping us to refine in-cabin software before a human tester ever encounters it on a touchscreen or dashboard display. This automated testing helps GM engineers detect software bugs before the code is integrated into vehicles, allowing us to identify issues – and implement solutions earlier in the development cycle.
After simulated testing is complete, engineers drive vehicles running our newest software builds at Milford Proving Ground in Michigan and other test sites. Some of these company-owned vehicles are equipped with a custom device, developed by a GM engineer, that runs automated software tests while the vehicle is not being driven. When those cars are parked overnight, machine learning-powered testing takes over to help enable a more predictable outcome at the time of release.
With automation and simulation, we can run these tests in a fraction of the time that physical testing would require. Virtual testing is always backed up by real-world data from GM employees who test-drive new vehicles across the U.S. These “four corners” drives aim to make sure our software is ready for environments drivers are most likely to encounter.
GM operates software testing labs in Michigan, California and Ontario, Canada. In one Michigan facility alone, more than 300 test benches are dedicated to finding bugs and malfunctions in dashboard infotainment systems. Thanks to AI and machine learning, we’re running more intelligent tests every hour of every day.
SOURCE: GM
