Modeling of a Soft Robotic Tongue Driven by Compressed Air

Abstract

A novel soft robotic tongue is proposed for the purpose of presenting typical tongue movements in speaking, which is pneumatically actuated and supposed to be applied in pronunciation teaching device and humanoid robot to gain verisimilitude. The actuation and control system is mainly composed of solenoid proportional valves (ITV0010, SMC, Japan), DAQ card (PCIe6321, NI, USA) and motion tracking system (OptiTrack, NaturalPoint, USA). A theoretical model based on continuum mechanics is established building the relationship between input air pressure and output deformation. On the basis of the proposed theoretical model and Artificial Neural Network (ANN), an integrated control algorithm is preliminarily raised for different purpose of use in static and dynamic control. A series of verification experiments are carried out based on OptiTrack system in order to test motion controllability and working performance of the robotic tongue. The experimental results indicate that the proposed theoretical model is more suitable for static shape control due to obvious deformation delay compared with input air pressure. High deformation repeatability guarantees the precision of ANN based control method which is more suitable for dynamic motion control. The results of performance test show that actual movements of the robotic tongue generally meet the demand in demonstrating tongue shape when speaking.