Fastbot: A Multi-Agent Model-Based Test Generation System : Beijing Bytedance Network Technology Co., Ltd.

Model-based test (MBT) generation techniques for automated GUI testing are of great value for app testing. Existing GUI model-based testing tools may fall into cyclic operations and run out of resources, when applied to apps with industrial complexity and scalability. In this work, we present a multi-agent GUI MBT system named Fastbot. Fastbot performs model construction on the server end. It applies multi-agent collaboration mechanism to speed up the model construction procedure. The proposed approach was applied on more than 20 applications from Bytedance with more than 1500 million monthly active users. Higher code coverage in less testing time is achieved with comparison of three other automated testing tools including Droidbot, Humanoid and Android Monkey.

[1]  Demis Hassabis,et al.  Mastering the game of Go with deep neural networks and tree search , 2016, Nature.

[2]  Yuanchun Li,et al.  DroidBot: A Lightweight UI-Guided Test Input Generator for Android , 2017, 2017 IEEE/ACM 39th International Conference on Software Engineering Companion (ICSE-C).

[3]  Yue Jia,et al.  Sapienz: multi-objective automated testing for Android applications , 2016, ISSTA.

[4]  Alessandro Lazaric,et al.  Upper-Confidence-Bound Algorithms for Active Learning in Multi-Armed Bandits , 2011, ALT.

[5]  Yao Guo,et al.  Humanoid: A Deep Learning-Based Approach to Automated Black-box Android App Testing , 2019, 2019 34th IEEE/ACM International Conference on Automated Software Engineering (ASE).

[6]  Andreas Zeller,et al.  DroidMate: A Robust and Extensible Test Generator for Android , 2016, 2016 IEEE/ACM International Conference on Mobile Software Engineering and Systems (MOBILESoft).

[7]  Yuanchun Li,et al.  A Deep Learning based Approach to Automated Android App Testing , 2019, ArXiv.

[8]  Mayur Naik,et al.  Dynodroid: an input generation system for Android apps , 2013, ESEC/FSE 2013.