The autonomous car of the future needs to be able to avoid malfunctions turning into disasters and a Pittsburgh-based company, Edge Case Research, has developed a robotic platform that shows how a system reacts to faults and accidents.
The company, which has 25 clients including Toyota, runs an autonomous car through a set of tests aimed at breaking the autonomous algorithms. It turns off sensors, feeds wrong visuals, and changes the code to see if the system can adapt and avoid an accident.
See Also: Google patents pickup location finder for self-driving cars
The researchers expect malfunctions to be the major issue for auto and tech firms, although remote attacks are another potential threat.
This type of independent verification is necessary in the autonomous age, as a simple flaw in the system can cause fatal damage on the road. Edge Case Research says that most auto and tech firms are not spending enough time ensuring that their systems are secure.
A cultural disconnect?
“We definitely have a cultural disconnect. The folks in the robotics world don’t necessarily think about these kinds of issues. They are more concerned in building the right kind of algorithm,” said Mike Wagner, co-founder of Edge Case Research to the Tribune-Review. “Right on the heels of it, when you’re ready to deploy it, safety engineering says you need to test the robustness of it. You have to test the fault tolerance of it.”
Edge Case Research has seen an uptick in interest from auto and tech firms, as we move ever closer to a fully autonomous age. Some automakers are worried that a Tesla crash disaster could lower consumer confidence in self-driving and lead to heavier regulation.
Google’s self-driving division Waymo has revealed the way it will secure its own cars, by moving most of the computing into the car. The self-driving vehicles will remain mostly offline throughout the journey, to keep remote hackers at bay.