In-Situ Classification of Soil Types Exploiting Electrical Impedance Tomography with a Robotic Actuating Probe

Abstract

Soil is a vital resource for various industries, including agriculture, engineering, and manufacturing, where accurate in-situ classification is essential for a wide range of applications. Electrical Impedance Tomography (EIT) enables real-time soil classification by capturing complex impedance data across varying distances. This study presents a novel approach integrating EIT with actuating probes to dynamically generate rich datasets for distinguishing soil types and moisture levels. By utilizing eight moving electrodes multiplexed across 32 channels, this system overcomes the limitations of traditional laboratory-based methods, such as time constraints and data skew caused by non-homogeneous inclusions. The moving electrode design significantly outperforms the stationary setup by 21%, achieving an average classification accuracy of 93% across varying moisture levels of sand, clay, and silt combinations. Experimental results on a larger data set demonstrates a classification accuracy of up to 79.7% across 25 different soil-moisture combinations, underscoring the technique’s potential for effective in-field soil analysis The improved accuracy achieved through actuation, compared to stationary probes, suggests broader applications in precision agriculture, civil engineering, and environmental monitoring.