Songyang Yan Autonomous driving systems testing

Research

Testing autonomous driving systems through simulation, scenario generation, and repair

My research asks how autonomous driving systems fail, how those failures can be reproduced in simulation, and how test infrastructure can make safety evaluation more systematic.

Research Themes

Autonomous driving systems testing

I approach this area with reproducible simulation workflows, scenario-level evidence, and engineering artifacts that can be inspected and extended.

Simulation-based safety evaluation

I approach this area with reproducible simulation workflows, scenario-level evidence, and engineering artifacts that can be inspected and extended.

Scenario generation and coverage

I approach this area with reproducible simulation workflows, scenario-level evidence, and engineering artifacts that can be inspected and extended.

Closed-loop co-simulation

I approach this area with reproducible simulation workflows, scenario-level evidence, and engineering artifacts that can be inspected and extended.

Configuration bug localization and repair

I approach this area with reproducible simulation workflows, scenario-level evidence, and engineering artifacts that can be inspected and extended.

Research Position

From empirical failures to reusable evaluation infrastructure

The goal is not only to find individual failures, but to build test methods, benchmarks, and toolchains that make those failures easier to explain, reproduce, and prevent.