Electrical capacitance tomography with a robot gripper for liquid level measurement with machine learning

Purpose Accurate liquid level measurement in long horizontal production pipelines is essential due to their complex flow dynamics, varying diameters and the need for multiple measurement points. This study aims to introduce a novel approach to using electrical capacitance tomography (ECT) integrated with a robot gripper. Design/methodology/approach The proposed system features two capacitive tactile sensors mounted on each finger of a robot gripper for liquid level measurement. Experiments were conducted at different liquid levels and various electrode angles relative to the liquid surface. Findings Experimental results indicate that measurements are reliable within a maximum tolerance angle of ±45°. By applying dimensionality reduction through machine learning algorithms, the computation time is significantly reduced, and a classification accuracy of 96% is achieved with seven liquid levels. Originality/value This study presents a nonintrusive, adaptable and movable liquid level measurement method suitable for long pipelines of varying diameters and multiple measurement points. The findings contribute to enhancing product quality, minimizing waste and optimizing resource utilization in beverage and liquid food manufacturing.