What Software Engineers Get Wrong About AV Safety
Safety in autonomous driving isn't just about uptime and test coverage.
Industrial PhD Student ยท Volvo Cars & Chalmers University of Technology
I research safety in autonomous driving, with a focus on runtime monitoring using out-of-distribution detection. My work aims to answer a hard question: how do we know when an autonomous vehicle's perception system is about to fail โ before it does?
Safety in autonomous driving isn't just about uptime and test coverage.
I'm an Industrial PhD student jointly at Volvo Cars and Chalmers University of Technology, researching safety in autonomous driving with a focus on runtime monitoring and out-of-distribution detection.
My path here was anything but linear. I started as a lover of design โ drawn to how technology shapes human experience. That curiosity led me to study Computer Science at the University of Ghana, then Information Technology at Carnegie Mellon University as a MasterCard Foundation Scholar, where I initially planned to pivot into HCI.
A summer at EPFL's Mathis Lab โ building ML-powered tools for neurological patient assessments โ showed me how deeply software can affect real lives. A role at Volvo Cars then pulled me into the automotive world, and I never looked back. Today I'm working on one of the hardest open problems in AV safety: detecting when a perception system is about to fail, before it does.
๐ง etornam@chalmers.se (academic)
๐ง emmanuella.ametsi@volvocars.com (professional)