Web applications support a wide range of activities nowadays, from e-commerce to health management, and it is important to ensure their quality. Functional testing of modern single page web applications is difficult because of their dynamic nature and heterogeneity. Most of the existing approaches resort to web crawling to explore the dynamic states of a web application. Crawlers automatically extract a navigation model of the web application and generate test cases from it, but they lack domain knowledge, thus they might not be able to exercise significant parts of the web application. In my thesis work, we take advantage of the navigation model implicitly specified by developers when they use the Page Object design pattern for web testing and we propose two novel, respectively search based and diversity based, techniques that take into account the navigation structure to generate end-to-end (E222E) test cases.