STADA: Specification-based Testing for Autonomous Driving Agents
Published in IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS 2026), 2026
STADA systematically generates the space of scenarios defined by a formal specification expressed in temporal logic (LTLf), constructing initial scenes and diverse scenario continuations directly from formal requirements rather than relying on manual or randomly sampled scenario generation. Across our evaluation, STADA achieves more than 2x higher coverage than the best baseline on the finest-grained coverage criteria while requiring significantly fewer simulations. The approach generalizes beyond autonomous driving to other simulation-rich testing domains.
Recommended citation: Joy Saha, Trey Woodlief, Sebastian Elbaum, Matthew B. Dwyer. "STADA: Specification-based Testing for Autonomous Driving Agents." IROS, 2026.
Download Paper