In this talk I will give an overview of the results achieved by a recent collaborative project called IV4XR, which stands for Intelligent Verification and Validation of Extended Reality (XR) Systems. Much of the research in the project has focused on automated game testing which is deemed as a crucial stepping stone towards verification and validation of XR systems in general. Computer games, or XR systems in general, take a lot of effort to build and the market is very competitive. While Quality Assurance is understood to be important, the Industry lacks automated testing technology and consequently have to rely on time consuming and expensive manual play testing. Moreover, for such a system user experience is often an important aspect, for which there exists no clear idea how its assessment can be automated. The problem is not just the lack of tools. It is also fundamental.

Computer game is a very challenging domain for software testing due to its very fine grained interaction space, limited observability, non-determinism, and even hostile elements that we do not usually have when testing traditional applications. All these make them untractable to be directly targeted by existing automated testing algorithms such as search based tetsing (SBT), model based testing (MBT), or machine learning (ML) based testing. In this talk I will show our novel agentbased approach that underlies IV4XR, with tactical and goal-based programming. This allows more resource efficient reasoning-based test algorithms to be implemented rather than relying on brute force approaches. We will show how the approach can be combined with a computational model of emotion to also cover user experience testing. We will also show how such an agent-based approach can provide a middle layer that enables other automation techniques such as SBT and ML to much more efficiently target computer games. At the end we will discuss how to proceed to cover other types of XR systems.