The growing Indian population, post-harvest losses and the ever-increasing demand of high quality, export-grade products which are produced under stringent safety protocols and standards highlights the need to establish stringent, swift, and objective quality evaluation procedures to food and agricultural products. The aim of this research was to develop a smart fruit sorting machine that would distinguish healthy and flawed guavas. The device is made up of a conveyor belt, a separator, an optical sensor array, and an Arduino Uno microcontroller board. Based on image-analysis algorithms that were run on an embedded system architecture. The experiment was run on a field-programmable array which was implemented on a programmable logic controller, running machine-learning algorithms on Matlab and coordinating the functionality of the relay with the use of MATLAB. The system was experimentally tested and proved accurate by physical sorting and checking 200 samples, with an accuracy of 97% of healthy and damaged fruits being identified based on spectral differentiation in the conveyor-segregator subsystem.