Stuttgart, 2025. A mechanic plugs a laptop into a car to update its software. But instead of a one-click job, he gets a cascade of errors. Every electronic control unit in the car, whether the brake system or the infotainment, runs its own “dialect.” They’re made by different companies and tied to a specific chip, like a key to a lock. To get them to work together and accept updates, you need a universal interpreter — called a hardware abstraction layer (HAL). Think of it as a diplomatic protocol: it translates from the language of computer hardware to the language of software.
Researchers from Stuttgart looked at other fields that already use such interpreters: smartphones (so apps run on any phone), industrial robots (for flexible factory setups), and network equipment (which must adapt quickly). They compared over a dozen automotive interpreters against strict criteria: how fast they respond to danger (instant reaction), whether they can support self-learning features like traffic sign recognition, and how easy it is to port the software to another chip.
The results split the interpreters into two camps. Some work like a soundproofed negotiation room: they fully isolate vital parts (steering, brakes) from the rest, preventing any error from compromising safety. Such systems use almost no resources: for instance, one of them, QNX, uses less than 2% of the processor’s power. Others are like an open forum: they let different programs talk freely, but without strict oversight. This is handy for cameras and radars that transmit images, or for algorithms that learn from mistakes. But for critical systems, such flexibility is dangerous.
So the researchers proposed a hybrid approach — a diplomatic pact. Safety-critical nodes (brakes, steering) are placed in isolated rooms with strict access control. Everything else (navigation, music, climate control) communicates via a universal protocol, much like web services. That way, a glitch in the entertainment system won’t affect driving, and a new version of the navigation software can be loaded over the air without reflashing every module. This idea has been pondered since the days of von Neumann, but now it’s hitting the road.
The Stuttgart mechanic plugs in the laptop again. This time, the car swallows the code instantly, as if it understands commands by intuition. It’s thanks to these interpreters that millions of cars will either become full citizens of the digital world or remain a heap of uncomprehending metal.
QNX spends less than 2% of processor time on its own overhead, like an interpreter at a high-stakes meeting who gets distracted for just a couple of seconds per hour.
In the TV series Knight Rider, the car KITT could get new abilities over the air. Today’s work brings that fiction closer to the factory floor.