Analyzing Vibration in Omni-Wheels: A Design of Experiments Approach to Optimizing Omni-Wheel Selection

Abstract

Omni-wheels, known for enabling holonomic motion in robotic systems, often introduce vibration due to their complex geometry and multiple contact points. Unlike caster wheels with established testing standards, omni-wheels lack comprehensive characterization methods. While parallel studies by Ilkbahar [1] and Donnellan [2] explore their rolling resistance and static load capacity, a systematic analysis of vibration characteristics remains absent from the literature. This thesis presents an investigation of the vibration behavior of various omniwheel designs using a Design of Experiments (DOE) approach. A full factorial experimental design was developed, considering factors such as wheel type, rotational speed, applied load, and wheel orientation angle. Individual regression models were developed for each of six wheel types, treating operational parameters as continuous variables. Vibration levels were measured using root mean square (RMS) acceleration, derived from Fast Fourier Transform (FFT) and Power Spectral Density (PSD) analyses of accelerometer data. Results show that rotational speed consistently increased vibration across all wheel designs, while lateral motion (90° angle) consistently reduced vibration compared to forward motion. The effect of applied load varied significantly between wheel designs, with some wheels showing reduced vibration under load while others remained unaffected. Wheels DZ(1) and Vex(5) demonstrated the lowest average vibration levels, though post-test inspection revealed trade-offs with durability, including roller deformation and material degradation. Interaction effects, particularly between angle and speed, were statistically significant for all wheel types, indicating that the benefits of lateral motion are enhanced at higher speeds. This research provides a framework for optimizing omni-wheel selection to minimize vibration by developing wheel-specific predictive models that quantify sensitivities and interaction effects across various designs and conditions, improving system performance and stability. The findings highlight that wheel selection must consider not only vibration performance but also trade-offs with durability and rolling resistance, establishing vibration characteristics as a critical consideration alongside other performance metrics when selecting omni-wheels.