A Robot and Sensor Integration Method to Measure Fatty Acid Composition in Salmon Fillets

Summary

In this work, we present a robot-sensor integration solution to allow in-line estimation of omega-3 fatty acid features in salmon fillets through Raman spectroscopy technology. The composition of omega-3 fatty acids in salmon fat is a critical food quality indicator for the salmon farming industry, as a high content contributes to a redder fillet and fewer unwanted dark spots and, consequently, leads to a higher price tag. Nowadays, Raman spectroscopy is conducted manually using a commercial all-in-one Raman system for rapid measurements in a laboratory or close to the production line. In general, this activity can be tedious, error-prone, time-consuming, and labor-intensive. To address such challenges, we propose an autonomous robotic Raman spectroscopy system synergized with a computer vision system, an AI-based learning approach, and a motion planning algorithm. We then developed a proof-of-concept demo using a Raman sensor mounted ad-hoc on a 6-DoF robot arm, an RGB-D camera, a conveyor system, and printed/natural salmon fillets, emulating a lab-scale industrial environment. All software architecture is implemented using the ROS framework and open-source Python libraries. Our robot-sensor integration solution can provide promising results, as the robotic system allows for faster and more accurate measurements of omega-3 fatty acid content in salmon fillets moving over the conveyor belt at low and high speeds, improving operational efficiency, and scaling the business.