Chuchu Fan's research focuses on developing novel ways to automate the design of safe, reliable, and robust decision and control in autonomous systems – for systems ranging from aerial vehicles and spacecraft to robots that zip along and navigate among pedestrians on sidewalks and streets. Her research endeavors are organized along with three directions, which together create a systematic framework of understanding theoretical, algorithmic, and applicable aspects of building dependable and robust AI-powered large-scale autonomous systems.
Her solutions bring together numerical simulation data, symbolic sensitivity analysis of the physics-based components, and core approaches from software verification like equivalence checking and fixpoint analysis.
Chuchu has invented game-changing algorithms that embody fundamental new ideas for practical and rigorous safety verification of autonomous systems, and her software implementations have shown for the first time how complex autonomous control systems could be automatically checked for safety. Research labs and at least two companies are adopting these technologies.
Her research goal is to fundamentally transform the conventional trial-and-error paradigm and develop computationally efficient formal techniques for designing and analyzing real-world AS by providing high coverage at a low cost.
Chuchu's work advances the
mathematical and algorithmic underpinnings for the provably safe and correct
design of AS, resulting in rigorous approaches that can generate provably
correct decision systems, provide safety guarantees, and perform root-cause
analyses to discover bugs at the early design stage. She develops algorithms
and software tools that can build the first line of defense against design
defects making their way into unsafe autonomous systems. Her techniques can
help engineers build reliable devices in a variety of application domains like
cars, planes, satellites, medical devices, and robots, and save lives and
billions of dollars in costly recalls.