Automated robotic arm system for real-time multi-parameter quality assessment of raw milk in cheese manufacturing

Abstract

Ensuring the quality of raw milk is critical for consistent cheese manufacturing, yet traditional laboratory-based testing methods are slow, labor-intensive, and impractical for decentralized rural supply chains. This study presents a portable, fully automated robotic arm system for real-time, multi-parameter milk quality assessment. The system integrates pH, total dissolved solids (TDS), temperature, density, and color sensors into a single testing cycle of under five minutes. A four-degree-of-freedom robotic arm ensures precise and repeatable probe positioning, reducing contamination and accommodating varied container types. An AI-based Support Vector Machine (SVM) classifier, trained on multi-sensor data, achieved 97.1% classification accuracy, outperforming static threshold logic, particularly in borderline cases. Environmental control features, including an LED-based optical chamber and temperature-compensated TDS readings, improved robustness in non-climate-controlled rural conditions. Laboratory tests showed high agreement with ISO-calibrated references for pH and TDS. Field trials at rural milk collection centers in Sri Lanka demonstrated over 96.5% agreement with laboratory classifications. Although individual sensor readings (e.g., pH, temperature) are rapid, the integration of automated handling, sensor switching, and AI-driven classification reduced total testing time per sample by approximately 35% compared to manual workflows. The modular design allows for scalability, easy maintenance, and adaptability to resource-limited environments. By enabling rapid, non-destructive, and chemical-free testing, the system addresses critical challenges in rural dairy networks, improving decision-making, reducing spoilage risks, and supporting higher quality assurance standards in cheese production workflows.