Self-driving cars can be fooled by a surprisingly simple pattern

A simple pattern of black-and-white stripes might be enough to trick an autonomous vehicle into misjudging how far away an obstacle is.

New research from the University of Florida (UF) has exposed this overlooked weakness in how self-driving cars perceive depth.

How stereo cameras get confused The study zeroes in on autonomous systems that rely on stereo cameras.

Rampazzi’s team worked alongside two universities in Japan, testing a range of sensors, algorithms, and AI models.

That pattern fooled the autonomous vehicle’s perception system into thinking part of the car ahead was closer than it was.