Fastbot Automated Testing: Enhancing iOS App Stability Through Intelligent Testing

Fastbot is an intelligent automated testing service developed by Quality Lab and GIP iOS Platform Architecture Team, utilizing machine learning and reinforcement learning to improve test coverage and stability, achieving over 100% improvement in problem discovery and code coverage compared to traditional Monkey testing.

The article details three main areas of Fastbot implementation: general capability development (visual operation, screen freeze detection), capability expansion (business customization, UI detection, keyboard optimization), and data consumption (offline qualification processes, test machine allocation strategies).

Key enhancements include scene limitation functionality for deep ViewController coverage, accessibility detection using Google's GTXLib to identify seven categories of accessibility issues, and task exception recovery mechanisms including screen freeze detection and keyboard scenario optimization.

The offline qualification process consists of three parts: quality assessment, exception consumption, and release checkpoints, helping teams make informed release decisions. Current results show 45.76% problem recall rate and 80% ViewController coverage.

Future plans include improving traversal capabilities, expanding coverage metrics, integrating offline recording playback, and enhancing UI assertions and performance testing capabilities.