To move or not to move? Analyzing motion cueing in vehicle simulators by means of massive simulations

Abstract

Motion platforms and motion cueing algorithms (MCA) have been included in virtual reality applications for several decades. They are necessary to provide suitable inertial cues in vehicle simulators. However, the great number of operational constraints that these devices and algorithms suffer, namely limited physical space, elevated costs, absence of sufficient power, difficulty of tuning and lack of standardized assessment methods, have hindered their widespread use. This work tries to clarify open questions in the field, such as: How important is MCA tuning? How much does size, number of DOF and power/latency matter? Can the absence of motion be better than poor motion cueing? What are the key factors that should be addressed to enhance the design of these devices? Although absolute certain answers cannot be given, this paper tries to clarify these research questions by performing massive experiments with simulated motion platforms of different types, sizes and powers. The information obtained from these experiments will be important to customize the design of real devices for this particular use. Ideally, subjective experiments with human experts would have been preferred. However, the use of simulated devices allows comparing many different motion platforms. In this paper, forty of these devices are simulated, optimized by means of a heuristic algorithm and compared with objective indicators in order to measure their relative performance using the classical MCA, something that would require an unreasonable amount of effort with real users and real devices. The obtained results show that MCA tuning is of the utmost importance in motion cueing. They also suggest that high power can usually compensate for lack of size and that a 6-DOF motion platform slightly improves the performance of a 3-DOF motion platform.