Abstract: With the development of new generation mobile communication technology and chips, the number of    intelligent mobile terminal users is increasing. In order to quickly seize the market, developers have shortened the development cycle of intelligent terminals, which raises higher requirements for the reliability and stability of application systems. Automation testing technology is an important means to ensure the high reliability and strong stability of these intelligent terminals. This paper discusses the black box testing technology and white box testing technology of Android system respectively, combined with the architectural characteristics and component features of mainstream intelligent terminals. In terms of black box testing, this paper compares and analyzes the latest UI testing and fuzz testing technology and tool usage, and evaluates their effects in ensuring the reliability and stability of application systems. In terms of white box testing, this paper summarizes the technology of automatically generating test cases, dynamic and static taint analysis technology, third-party library detection technology, and permission detection technology. Finally, with the emergence of emerging technologies such as AI models, more and more intelligent terminal devices are starting to carry various deep learning models. The opacity of these models makes the internal decision-making process difficult to explain and understand, so the black box testing is increasingly important in evaluating model reliability and stability. Automation testing is undergoing a transformation from traditional rule-based testing to more intelligent machine learning-driven testing. In the future, it is necessary to introduce emerging technologies such as AI models into existing intelligent terminal testing practices, which has become a necessary trend to solve this problem.

Key words: intelligent terminal, Android system, software testing, search-based test case generation

摘要： 随着新一代移动通信技术和芯片的发展，智能移动终端用户规模不断增加。为了快速抢占市场，开发商缩短了智能终端的开发周期，这对应用系统的可靠性、稳定性等提出了更高的要求，而自动化测试技术是保障这些智能终端高可靠、强稳定运行的重要手段。结合目前主流智能终端的架构特点和组件特征，分别探讨了安卓系统的黑盒测试技术和白盒测试技术。在黑盒测试方面，比较分析了最新的用户界面测试和模糊测试技术以及工具使用情况，评价了它们在保证应用系统可靠性和稳定性方面的效果。在白盒测试方面，总结了自动生成测试用例技术、动静态的污点分析技术、第三方库检测技术和权限检测技术。随着人工智能大模型等新兴技术不断涌现，越来越多的智能终端设备开始搭载各种深度学习模型，这些模型的不透明性使得内部决策过程难以解释和理解，从而黑盒测试在评估模型可靠性和稳定性过程中越发重要。自动化测试正在面临从传统规则基础下的测试向更加智能的机器学习驱动的测试转变。未来将人工智能大模型等新兴技术引入到现有的智能终端测试实践中，成为解决这一问题的必然趋势。

关键词: 智能终端, 安卓系统, 软件测试, 基于搜索的测试生成