This Ph.D. thesis investigates how interactions with autonomous systems can be simulated to explore a wide range of scenarios before autonomous driving technology is deployed to the public. The advent of autonomous vehicles creates new challenges for driving simulation: user experience, non-driving related tasks, and driver-driver interactions become more relevant. Given the new challenges emerging intelligent systems bring, this dissertation develops three new simulator systems that allow interactions to unfold more freely than traditional implementations. This allows researchers to discover how and which design choices matter for seamless interaction and safe introduction to the public. The first simulator explores how interactions between a driver/passenger and an autonomous vehicle unfold, especially in critical traffic situations; the second extended this concept to include real-world traffic. The third simulator enabled the examination of the interaction between traffic participants, specifically driver-driver interactions, and their strategies to resolve ambiguous traffic situations. Besides enabling immersive, replicable, and reusable research, the simulators are designed to capture and reproduce rich data streams from participants' reactions. The unified view of these data streams facilitates the reconstruction of the interaction through qualitative behavioral analysis. The thesis concludes with an outlook on how these methods and simulators, in particular the discovery-based approach, could find applications within the research fields of Human-Robot Interaction and Human-Computer Interaction.