Mark Coletti
Papers
1
Total Citations
7
H-Index
1
About
Mark Coletti is a researcher whose work lies at the intersection of autonomous systems, evolutionary computation, and simulation-based validation. His key research areas include adversarial testing of autonomous vehicles, evolutionary algorithms for scenario discovery, and the development of robust evaluation frameworks for intelligent systems. Coletti’s major contribution is the innovative repurposing of adversarial evolutionary algorithms—specifically the Gremlin system—to diagnose and improve driving quality evaluation criteria for autonomous vehicles. Rather than simply finding scenarios where an autonomous model fails, his approach proactively identifies weaknesses in the criteria used to judge driving performance, enabling more rigorous and reliable testing. This work, published in 2021 and garnering 7 citations, demonstrates a novel methodology for stress-testing evaluation metrics using a simulated "perfect driver" robot in a virtual town environment. Coletti’s research is notable for shifting the focus from merely detecting failures to improving the very standards by which autonomous driving is assessed, offering a more foundational path toward safer and more trustworthy autonomous systems.
Research Focus
Key Achievements
Top Papers
- 1