3D Vision System for Dairy Cow Tail and Teat Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Dairy milking operations face inefficiencies due to the variability in cow positioning and teat placement, making it difficult for robotic arms to accurately and efficiently perform tasks without hard coding specific movements and positions, and the presence of cow features like tails can hinder operation accuracy.
Innovation Solution
A vision system utilizing a 3D camera and processor to capture and analyze 3D images of dairy livestock, enabling real-time detection and positioning of legs, teats, and tails, allowing the robotic arm to adjust its movements and avoid obstacles, thereby improving accuracy and speed without the need for hard coding.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If a robotic arm is used for automated milking operations, then productivity increases, but the system fails to adapt to variable cow positioning and teat placement without hard-coded movements
Solution Approach 1:
The vision system performs preliminary detection and mapping of cow anatomy (teats, legs, tail position) before the robotic arm begins its milking operation. This advance information gathering allows the robotic arm to adapt its movements dynamically without requiring hard-coded positioning for each variable scenario.
Solution Approach 2:
The system continuously monitors cow movement and teat position during milking operations using the vision system, providing real-time feedback that allows the robotic arm to adjust its movements dynamically. This closed-loop control enables adaptation to positioning variability while maintaining high productivity.
2Device complexity
If the robotic arm operates without real-time detection capabilities, then device complexity is reduced, but measurement precision of teat position and cow movement deteriorates
Solution Approach 1:
The vision system acts as an intermediary between the robotic arm and the cow, providing precise measurement data about teat position and cow movement without requiring complex sensors directly on the robotic arm. This separates the detection function from the execution function, maintaining relative system simplicity while achieving high measurement precision.
3Productivity
If the robotic arm moves quickly through milking operations, then productivity increases, but accuracy of operation deteriorates due to cow movement and tail obstruction
Solution Approach 1:
The vision system performs preliminary detection of the tail position and anatomical features before the robotic arm executes its milking sequence. This advance knowledge allows the system to plan accurate trajectories that account for potential obstructions, maintaining precision even at higher operating speeds.
Solution Approach 2:
The system transitions from static, hard-coded movement paths to dynamic, real-time trajectory adjustment based on vision system feedback. This allows the robotic arm to maintain high speed while adapting its path dynamically to avoid tail obstruction and accommodate cow movement, preserving operation accuracy.
4Device complexity
If hard-coded movements are used for robotic arm positioning, then device complexity is reduced, but adaptability to different cow positions deteriorates
Solution Approach 1:
The patent replaces hard-coded mechanical positioning control with a vision-based detection and control system. The vision system captures images, identifies anatomical features, and generates dynamic positioning commands, substituting rigid mechanical control logic with flexible optical sensing and software-based adaptation.
Data Source
AI summary
A system that includes a three-dimensional (3D) camera configured to capture a 3D image of a rearview of a dairy livestock in a stall and a processor. The processor is configured to obtain the 3D image, identify one or more regions within the 3D image comprising depth values greater than a depth value threshold, and s to identify a thigh gap region from the one or more regions. The processor is further configured to demarcate an access region within the thigh gap region and demarcate a tail detection region. The processor is further configured to identify one or more tail candidates within the tail detection region, to identify a tail candidate that corresponds with a tail model as the tail, and to determine position information for the tail.