Self-Shape Sensing Soft Pneumatic Grasper Based on Piecewise Liquid Metal Sensor and Piecewise Variational Curvature Model

The shape estimation function can help solve end positioning or gripping control of soft robots. However, there is a lack of sensing and modeling techniques for accurate deformation estimation and soft robots with axial elongation, e.g., pneumatic soft robotic graspers. This paper presents a self-shape sensing pneumatic soft grasper with integrated liquid metal composite piecewise curvature sensors. Ga-Ln-Sn alloy was used as the basis of the sensor with addition of NdFeB and Ni. Then, a piecewise variable curvature model was developed to predict the deformation of the robotic fingers. A three-fingers soft robotic grasper (working similarly as a two-finger grasper on the 2D working plane) was built to test the performances of the sensor and the model. The result indicated that the grasper is not only capable of self-shape sensing, but also contact detection and gripping object size estimation. By statistical analysis, it is proven valid that the data collected by the sensor is able to go through machine learning processes to achieve gripping object shape identification.