Kodiak stands as an exception. Earlier this year, the company showed how one of its trucks reacted when an ethernet cable that supplied information to its self-driving system was cut. In that case, the Kodiak truck pulled to the shoulder along Interstate 45. It marked the first time a self-driving truck company demonstrated such capability; competitors like Aurora soon followed.
The ability to respond to a tire blowout, a cut cable or dozens of other potential malfunctions starts with the architecture of the technology's fallback systems.
Kodiak's self-driving system monitors for such faults 10 times per second. Engineers built the controller, which is responsible for actuating the truck and executing path planning, on a custom safety computer built to automotive-grade standards. Each Kodiak truck carries two controllers for redundancy.
Fallback systems should rarely be used but always be available, says Kodiak CEO Don Burnette.
"We can't control the hazards trucks will face on the open road, but we can control how the trucks behave when a critical situation occurs," he said.
Understanding how trucks behave in blown-tire scenarios is difficult to simulate because of variability in conditions and payload, Burnette said. That makes data collected from closed-course testing valuable.
For example, having information from the real-world 35-mph test helped Kodiak more accurately simulate what would happen in a 65-mph test. The truck still maintains its lane of travel.