Sheep Hock Reposition Vision System

Summary

Background
An critical challenge to successfully fully automating processing tasks has been the variability between animals.  This has required highly sophisticated visioning systems so as to provide the appropriate cutting lines to robots and other fully automated systems.
Hence MLA has funded a series of R&D projects on the visioning technologies that drive automation systems for meat processing.
The primary use for these technologies was to analyse each carcase and calculate optimal cutting lines, so as to maximise the value of the carcase.
Technologies included 2D, 3D & CT x-ray based scanning, as well as camera based systems.
In some cases, these technologies were also used to measure eating quality parameters. Where the project has this as its primary focus, results are reported under Eating Quality.
Research - vision and laser
Commencing in 2004, the Sensing and Automation science group in Food Science Australia developed automatic vision and laser scanning systems to accurately detect the location and shape of beef and sheep carcass features, with an aim towards further development of systems to automate processing work task operations.
Also, during 2005, the Sensing and Automation science group completed a project funded by AMPC and MLA: “Investigation and evaluation of Sensors for Adaptation to the Meat Industry” (PRTEC.032) that outlined beef and sheep slaughter tasks suitable to the application and development of vision and laser sensing systems.
The “Development of vision and laser sensing systems suitable for beef and sheep slaughter tasks” (PRTEC.042) followed in the 2005-06 financial year. This project progressed further the above work to develop vision and laser sensing for beef and sheep slaughter tasks.
Stage 2 of the vision & laser sensing systems suitable for beef and sheep slaughter tasks progressed further the development of sensing systems for process task automation under project A.TEC.0051.
Project P.PSH.0479 carried out successful trials of the TYZX Deepsea V2 Development System, 3D camera in MAR’s workshops. The system was then installed as part of the Beef Hock Cutting Project at JBS Swift Dinmore and proved to be very successful.
Project A.TEC.0062 funded reserach into a sheep hock reposition visioning system.
P.PSH.0619 and P.PSH.0635 funded beef carcass 3D camera motion sensing trials.
Research - x-ray
Project A.TEC.0047 demonstrated the ability to use sparse data from a carcase sensor (in this instance x-ray images) to deform a computer model such that key carcase attributes (cutting locations) could be identified (including extrapolating to locations not within the X-ray images) and outputted to a Plant System.
The A.TEC.0058 "Development of X-ray Sensing Technology" project was based on two selected tasks related to beef carcase processing. These tasks were beef split saw and beef scribe saw.
The MLA/RTL co-funded project P.PSH.0238 investigated the following.
Using the existing RTL lamb x-ray measuring system, determine a relationship between x-ray image intensity and lamb carcass mean density.Test the relationship by x-raying samples, weighing the samples, and comparing the measured weight with that predicted using the intensity/density relationship.Using the intensity/density relationship and the existing primal cutting coordinate analysis, develop and demonstrate a basic system to predict the weight of the three primal sections prior to the separation of the carcass.
Research - general
Project A.TEC.0041 was aimed at completing the intermediate step of real-time object tracking required to successfully implement Telerobotics in meat slaughter processing.
In PRTEC.0040, an experimental trial was carried out on a beef carcass for the purpose of obtaining samples of hide/connective tissue/subcutaneous tissue from specific areas of the carcass for further testing and analysis and to determine the suitability of the different characteristics of the hide/carcass interface tissue for detection. The types of sensing methods to be applied for detection, and the number and location of samples were decided from the work carried out in milestone one.  The three potential sensing methods trialled were:Thermal ImagingLow Voltage Radio Spectrum TransmissibilityFluorescent Spectroscopy