Toward Immersive Self-Driving Simulations: Reports from a User Study across Six Platforms

As self-driving car technology matures, autonomous vehicle research is moving toward building more human-centric interfaces and accountable experiences. Driving simulators avoid many ethical and regulatory concerns about self-driving cars and play a key role in testing new interfaces or autonomous driving scenarios. However, apart from validity studies for manual driving simulation, the capabilities of driving simulators in replicating the experience of self-driving cars have not been widely investigated. In this paper, we build six self-driving simulation platforms with varying levels of visual and motion fidelities ranging from a screen-based in-lab simulator to the mixed-reality on-road simulator we propose. We compare the sense of presence and simulator sickness for each simulator composition, as well as its visual and motion fidelities with a user study. Our novel mixed-reality automotive driving simulator, named MAXIM, showed highest fidelity and presence. Our findings suggest how visual and motion configurations affect experience in autonomous driving simulators.