Knowledge-Based System for Autonomous Control of Intelligent Mastication Robots

Currently, the methods used to analyse and evaluate the properties of food typically involve human sensory panels. These methods have the advantage of producing realistic, in-vivo results however, due to the subjective nature of sensory evaluation, results obtained from different panel members can be inconsistent. This inherent variability in experimental outcomes can lead to difficulties when interpreting and comparing results. To overcome this, a robot that is capable of emulating human mastication has been developed. This robot is driven by a single motor and can perform a family of rhythmic chewing motions that approximate the trajectories of human molar teeth. However, the robot is unable to adapt its chewing behaviour to varying food properties during the mastication sequence. This has a negative impact on results as mastication is a complex dynamic process and hence, cannot be accurately modelled by a static system. The aim of this research is to develop a knowledge-based system (KBS) that allows the robot to learn how human behaviour varies during the mastication process. This system operates using data regarding changes in mastication parameters with respect to changing food properties. This data is analysed (via machine learning algorithms) to uncover relationships between food properties and the parameters of mastication. These relationships form the robot's knowledge of human masticatory behaviour and can be used as a form of autonomous control. As a result, the robot (in conjunction with the KBS) is able to autonomously emulate human mastication while producing both, realistic and consistent results.