Testing of Cyber-Physical Systems (CPS) is manifesting unprecedented challenges due to the tremendous complexity and interdisciplinary nature of CPS. Thus, CPS testing must be driven by automated, scalable, optimised, and intelligent solutions. This keynote will focus on the key results from such CPS testing solutions that we developed in the last several years. These solutions rely on model-based testing, machine learning, and search algorithms. The results from the application of these testing solutions in industrial case studies and real-world systems will be presented together with the future research directions.