- The paper presents a modular VR simulator unifying e-scooter, Segway, unicycle, and skateboard metaphors for low-risk, controlled human factors studies.
- It employs a rapidly reconfigurable hardware-software architecture with dense multimodal sensing and wireless data mapping via Unity.
- Preliminary experiments reveal distinct workload and control differences across vehicles, informing design and safety protocols for VR locomotion research.
MicroVRide: A Modular VR Micromobility Simulator for Multivehicle Interaction Study
Introduction
The paper presents the design, implementation, and preliminary evaluation of MicroVRide, a modular, four-in-one VR micromobility simulator capable of emulating e-scooters, Segways, electric unicycles, and one-wheeled skateboards within a single hardware-software architecture. The primary contribution is the unification of disparate micromobility vehicle metaphors and physicality within a reconfigurable, sensor-rich platform, enabling rigorous human factors and interaction studies under controlled, low-risk VR conditions. The simulator is positioned as a foundational infrastructure to catalyze both behavior-oriented studies and expanded VR locomotion research, especially concerning non-cycling micromobility modalities, which have remained underexplored in prior art.
System Architecture
MicroVRide constitutes an extensible modularity at both hardware and software layers to accommodate diverse vehicular control schemas. Its physical core is a standing platform that adapts to the requirements of each vehicle via the rapid exchange of peripheral components and sensor repositioning. The platform supports dense multimodal sensing: IMUs (to capture pitch, roll, yaw for steering and movement), force-sensitive resistors (for foot-pressure based inputs, essential for Segway-style metaphors), and optional handlebar and throttle units (for e-scooter and Segway modalities). Vehicle-specific input mapping is achieved through a wireless data pipeline interfacing with a Unity-driven VR application, where individual controller modules parameterize the virtual physics, constraints, and embodiment consistent with the real-world counterparts.
The e-scooter scenario leverages handlebar IMU data for yaw steering and thumb throttle for velocity. Segway control integrates foot pressure (forward/back) and handlebar roll (steering), reflecting real-world proportional control. One-wheeled skateboard and unicycle metaphors exploit full-body balancing and omni-directional tilt capabilities enabled by a hemispherical passive balance board, with IMU signals mapped to both speed and steering. This architecture prioritizes rapid context switching (≈1 minute hardware reconfiguration) and input fidelity tuned per vehicle, thereby supporting inter-condition studies without introducing systemic cross-vehicle artifacts.
Experimental Protocol
A within-subjects feasibility study (N=12) was conducted to investigate the experiential profile and input demands across the four configurations. The protocol comprised both training and task-driven exposure in VR, where participants navigated structurally-matched urban routes by collecting virtual coins. Control order was counterbalanced. The evaluation harnessed subjective metrics including raw NASA-TLX (assessing task workload across six dimensions), Likert ratings on controllability, stability/predictability, embodiment, and open-feedback interviews to triangulate on practical usability and user adaptation.
Numerical Results
The e-scooter condition consistently yielded the lowest perceived workload (mean NASA-TLX: 25.2, SD: 17.3) and highest ratings for naturalness and controllability, indicating strong transfer for riders with prior scooter experience. The Segway and one-wheeled skateboard presented moderate workloads (mean ≈50), but via distinct subscale profiles: Segway required higher sustained physical and mental effort, while the skateboard effect was dominated by physical demand with lower frustration. Notably, the electric unicycle resulted in the highest workload (mean NASA-TLX: 64.8, SD: 16.3) with pronounced spikes in perceived effort, performance pressure, and frustration. The latter configuration thus represents the upper bound of balancing and cognitive challenge in current VR micromobility simulation.
Control and stability ratings also stratified by vehicle. E-scooter and Segway modes demonstrated the highest subjective stability and user confidence, in line with their discrete affordances and wider real-world familiarity. Handlebar presence was frequently cited as a determinant of control transparency. The unicycle and skateboard demanded more lower-body and whole-body involvement, yielding lower control ratings but a distinctive sense of embodied interaction—critical for studies examining sensorimotor learning and adaptation.
Motion sickness was overall low, apart from two withdrawals (≈17%) due to discomfort. Exit interviews identified abrupt viewpoint changes on impact as the principal nausea trigger rather than steady-state VR locomotion, aligning with current theories on vestibular-visual conflict.
Theoretical and Practical Implications
MicroVRide fills a critical methodological void by supporting multimodal micromobility research with high input-fidelity and rapid vehicle interchange. It enables systematic comparison of different body-vehicle interaction metaphors on learning cost, workload distribution, and user adaptation under repeatable lab conditions—an advance over bespoke, single-vehicle systems.
From a practical perspective, the platform supports scalable, low-risk training environments for rider acclimatization to high-difficulty or unusual vehicles (particularly unicycles and one-wheeled skateboards) not widely accessible to the general public. The capacity to deploy rich logging across the full sensorimotor flow allows for post hoc behavioral modeling, safety studies, and potential integration with RL-based driver or motion prediction pipelines.
The findings reinforce the necessity for physical feedback, controller specificity, and real-time adaptation of safety protocols such as dynamic moderation of visual effects upon failure/collision to mitigate cybersickness in embodied VR. Future development directions include adaptive hardware ergonomics (handlebar/insole adjustability for body size variation), incorporation of haptic feedback, and closed-loop visual feedback overlays to facilitate faster skill acquisition and error recovery.
Theoretically, the results suggest that embodiment and control transparency are nontrivial to unify across divergent micromobility metaphors, and that user prior experience functions as a significant modulator of VR control intuitiveness and workload.
Conclusion
MicroVRide offers a validated, reusable platform for four distinct micromobility vehicles in VR, enabling human factors and embodied interaction research with high ecological and sensorimotor fidelity. Preliminary study results highlight strong vehicle-specific workload and control differentials, significant user adaptation, and a manageable cybersickness profile under controlled conditions. The modularity and extensibility of MicroVRide position it as an enabling tool for future research in VR-based micromobility safety, skill acquisition, behavioral modeling, and interface design.