To look at the tech news headlines, you might think that fully autonomous cars are just around the corner. While automakers aren’t yet ready to deliver a completely self-driving car to your local dealership, said Kerry Johnson, Automotive Product Manager for BlackBerry subsidiary QNX Software Systems, we are definitely on the road to making them a reality. In fact, he summarized data from a number of automakers and industry experts by saying that semi-autonomous cars will be available within the next 5 years and, by 2040, between 70% and 75% of all vehicles will be fully autonomous.
Johnson’s comments came during an Oct. 7 webinar, The role of a software platform when transitioning from ADAS to autonomous driving, which offered a solid look at where the automotive industry is today, where it is headed, and the challenges and opportunities that developers and automakers face along the way.
Self-driving cars are actually software- and sensor-driven vehicles, and it’s critical that the operating systems behind them are reliable, flexible and scalable across various hardware platforms and makes and models of cars, said Johnson. Because the point of a connected car is to connect with other systems, a hodgepodge of multiple closed systems that don’t interoperate will defeat the entire purpose of developing these vehicles.
(Read more about QNX and its leadership in the connected car platform market today.)
More importantly, Johnson said, connected car platforms must have security built in at the system level. Most of us have heard about vulnerabilities in automotive systems and, in truth, “connected cars have a broad attack surface,” he conceded. Weaknesses must be eliminated in connection points such as those within the vehicle (e.g., in the car’s internal sensor network), to other vehicles (e.g., braking when a sensor recognizes the car you’re following slows), to infrastructure (e.g., traffic lights and roads), and to cloud resources (e.g., weather and road condition alerts or software updates).
Engineering high availability is another key safety consideration in connected vehicles, Johnson said. The “five-nines” of availability, or 99.999% uptime accepted in most software development, means that a given system can tolerate five minutes of downtime per year. That might be acceptable for cloud storage, where lives and health aren’t at risk. But in a car moving on the highway at 100 kph, that 0.001% could be the difference between life or death if your car fails to brake — or brakes when it’s not supposed to.
“Still, we know that unforeseen faults will occur in vehicles,” said Johnson, so systems must be built using good software quality practices and architectures that enable the system to detect and recover from a failure before a dangerous (or even inconvenient) problem can occur.
Johnson explained that, increasingly, QNX is seeing from its customers a requirement for self-driving car software and systems to meet safety standards such as ISO 26262, which specifies the functional safety of electrical and electronic systems in road vehicles. Certification can also help ease consumer concerns about the safety of autonomous vehicles.
Johnson wrapped up the webinar by sharing lessons QNX has learned from its leadership in vehicle infotainment systems. The software integration challenges QNX has solved in car multimedia, navigation, human-machine interfaces (HMIs) and processors are the same types of challenges automakers will continue to face as the transition from today’s advanced driver assistance systems (ADAS) to tomorrow’s fully autonomous vehicles takes place.
Interested in learning more? Visit Techonline to view the archived presentation. To learn more about how QNX is advancing ADAS and self-driving vehicles, visit the QNX Auto Blog or the official QNX website.