A biomimetic tribological set-up was developed to better integrate physiological characteristics in oral friction studies. Polyvinyl alcohol was used to design tongue-mimicking samples (TMSs) with controlled bulk rigidity and surface roughness. The set-up allowed normal stresses and sequences of shearing motion to be applied between a TMS and a rectangular aluminum plate mimicking the hard palate. Cottage cheese with and without suspended microcrystalline cellulose particles of two sizes (small: 15 μm and large: 250 μm) was used as the model food in the experiments, making it possible to introduce heterogeneities well above the perception threshold, which often get excluded in conventional tribology studies. The feasibility of the set-up was tested by measuring how friction coefficient values were affected by food properties (viscosity, particle presence), TMS properties (surface roughness, bulk rigidity), and operational parameters (normal stress, shearing velocity). The measurements with the set-up were found to be repeatable, confirming its reliability. Particle presence led to increases in friction coefficient values, while food viscosity showed less of an influence, comparatively. Increases in the surface roughness and bulk rigidity of the TMSs led to a pronounced augmentation of the friction coefficient values. The friction coefficient values showed dependence on normal stress, increase in normal stress led to significant decrease in friction. Moreover, friction notably increased between the lowest and highest shearing velocities. In conclusion, this set-up proved to be efficient and reliable in measuring friction coefficient values and variation, paving the way for a better understanding of oral tribology and its dynamic nature.