FreneticV is a search-based testing tool based on an evolutionary approach that generates roads where an automated driving agent possibly fails the lane-keeping task. It uses a curvature-based road representation and, compared to its predecessor Frenetic, considers the validity of the generated roads. In particular, it tries to avoid generating roads with overly sharp turns, it detects self-intersecting roads, and has the capability of rotating and relocating roads so as to fit them in a given map.
FreneticV participated in the competition on testing Cyber-Physical Systems at SBST 2022.
- Code of FreneticV
- A preprint explaining FreneticV
- Ezequiel Castellano https://jp.linkedin.com/in/ezequiel-castellano-7076962b
- Stefan Klikovits https://klikovits.net/
- Ahmet Cetinkaya https://www.ahmet.ac/
- Paolo Arcaini http://group-mmm.org/~arcaini/