An Evaluation of Sensing Fat Depth for the Automation of Uniform Fat Trimming of Beef Striploin

An Evaluation of Sensing Fat Depth for the Automation of Uniform Fat Trimming of Beef Striploin

The trimming of excess fat from beef striploin primal is currently a manual process costing the Australian beef industry an estimated $89 million annually due to yield losses within beef processing plants. Robotics have been successfully deployed to address efficiency and productivity issues in similar products such as pork but are yet to be adapted to red meat. Specifically, a sensing technology capable of acquiring the fat depth information required for automated trimming is yet to be developed.

The work undertaken in this dissertation investigates the characteristics of the beef striploin primal and the processing considerations to develop a sensing performance framework for the application of beef striploin fat trimming. Computed-Tomography is used to provide a means of benchmarking the error present between manual measurements and the 'gold standard' of sensing technologies for this application in the performance metrics of accuracy (described as median error), precision (described as Inter-Quartile Range of error), linearity (described as R-squared quantity of actual vs predicted measurements), reliability (described as the expected probability of acquiring 'no read' measurements across surveyed nodes), and response time (described as the time required to acquire measurements). A weighted sensor performance evaluation framework was developed based upon analyses conducted on key aspects of the striploin primal fat profile and the fat specifications and operational constraints of the fat trimming process.

Fat depth measurement systems were developed using A-Mode and B-Mode ultrasound sensing technologies to obtain results that could be assessed using the developed weighted sensor evaluation framework. In applying this framework it was identified that the A-Mode (score: 47 / 75) ultrasound system was more suitable than B-Mode (score: 29 / 75) for ii implementation within a commercial automated fat trimming system. Though the majority of literature recommends the use of B-Mode ultrasound for fat depth measurements it was found that the performance metrics considered favoured simplicity and fast response typical of A-mode ultrasound technology.

Further work to validate the recommendation of A-Mode ultrasound technologies for uniform fat trimming of beef striploin is recommended by integrating this technology within an automated system for commercial use.

File reproduced in accordance with the copyright policy of the publisher/author/creator.