This study aims to investigate whether incorporating a model-based feedforward term into a position-controlled multi-chamber robotic chewing system can improve motion tracking consistency and whether such improvement is associated with more repeatable food fragmentation outcomes under controlled in vitro chewing conditions.
A dynamic model of the chewing mechanism is established and its consistency with measured motor torque is examined under in vitro chewing conditions. Based on this model, a velocity/acceleration feedforward controller is designed and integrated with a baseline proportional, integral, and derivative (PID) position controller. The effects of feedforward compensation are evaluated experimentally using repeated chewing tests on roasted peanuts (n = 5), with tracking performance quantified by trajectory root mean square error (RMSE) and trial-level tracking deviation (TrackDev) and fragmentation consistency assessed using particle size distribution (PSD) metrics and d50.
Experimental results show that feedforward compensation reduces trajectory RMSE by approximately 16%–17% across chambers and decreases PSD-root mean square of the bin-wise standard deviation (RMSD) by 23%–34%, while d50 changes remain chamber-dependent. Trial-level statistical analysis showed directionally consistent reductions in RMSE, TrackDev and PSDDev under feedforward compensation, with statistically significant TrackDev reductions in C1 and C3. Exploratory Spearman correlation analysis showed positive associations between tracking variability and PSDDev, particularly for TrackDev. These findings indicate that improved motion execution fidelity through physically grounded feedforward control was associated with reduced trial-to-trial variability in food fragmentation outcomes under the tested conditions.
This study demonstrates the application and experimental validation of model-based feedforward compensation in a multi-oral-chamber robotic chewing system. The originality lies in integrating an equivalent dynamic model with an existing position-controlled robotic masticator and evaluating how improved motion consistency is associated with the repeatability of in vitro food fragmentation outcomes.
