Development of intellectual support’s algorithm at manage decisions making in the feeding process based on a machine vision system’s on a farm

Anomalous feed overconsumption accounts for up to 10% and occurs in 5% of cases. This means that the farm loses an equivalent portion of milk yields and, consequently, profit. The feed volume can be controlled using a hardware-software system (HSS) that obtains spatial data through a 3D vision system, specifically a stereo camera, and analyzes it using segmentation methods, visual odometry, and volume calculation. The volume calculation involves summing the volumes of truncated parallelepipeds formed from the areas of sequential slices with equal width and varying height. The developed algorithm stitches adjacent images with an error of ±2 cm at a speed of 2.06 frames per second (at a movement speed along the feed table of 3 km/h). The error in determining the hay volume is, on average, 6.14%. Based on the comparison with the initial volume, judgments about the amount of residue are made: less than 3% is below acceptable, from 3% to 5% is satisfactory, and more than 5% is above acceptable. For the autonomous operation of the program, a prototype of the system's hardware part was developed-a box containing a stereo camera, a single-board computer, and a power supply.