Actuation planning and modeling of a soft swallowing robot

摘要

The paper presents a new methodology to solve the actuation and modelling problems of a soft-bodied swallowing robot (SR), developed for human swallow evaluation. To solve the actuation problem, a central pattern generator (CPG) based novel actuation scheme is developed and implemented to generate peristalsis in the robot. Machine learning based technique is used to determine the governing dynamics of the robot because presently the robot does not have any differential equation to describe its actuation principle or its physics. To profile and sense the peristaltic waveform, a flat version of the robot containing pneumatic chambers for actuation has been proposed to approximate the deformation of the original SR and the CPG actuation scheme is used to command the flat SR so that the pneumatic chambers can be inflated. The logic of actuation is motivated from the swallowing phenomenon in humans, have been implemented in real time. An optical motion detection system (Vicon) is used to track the displacement of the air chambers of the robot and hence, to generate time-series data for determining the governing differential equations of the robot by using l <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">1</sub> regularised machine learning technique. It is also concluded that the proposed method provides a promising new modelling technique for determining the governing dynamics of the robot where conventional modelling approaches are not applicable.