SoftSeparator: A Method for Dynamic Separation of the Signatures Generated by Two Cows Walking in Tandem over the Longitudinally-Split Floor of the StepMetrixTM

Title	SoftSeparator: A Method for Dynamic Separation of the Signatures Generated by Two Cows Walking in Tandem over the Longitudinally-Split Floor of the StepMetrixTM
Author	Dyer, Philip
Advisors	Tasch, Uri
Program	Engineering, Mechanical
UMBC Department	Mechanical Engineering
Document Type	thesis
Sponsors	University of Maryland, Baltimore County (UMBC)
Keywords	automation; separation; lameness; dairy; stepmetrix; Engineering, Mechanical (0548)
Date Issued	2007-12-04; *
Abstract	Due to a herd mentality and some styles of management, dairy cows will often bunch together when walking in dairy lanes. This bunching causes data loss for automatic walk-through devices placed in the lanes that require cows to pass through one at a time. The StepMetrixTM is such a device developed for early detection of hind-limb lameness in individual cows. The system utilizes a split-floor design with loadcells to record and analyze the ground reaction forces generated by a cow's walk after milking. The analysis uses a statistical model to produce an SMXTM score, which is associated with the probability of a cow being lame.; Upon examination of hundreds of data signatures, it was discovered that certain signatures have data generated by hind limbs that can be extracted to be used in the statistical lameness model. Specific types of steps produced by two cows walking in tandem were identified, and cutoffs were found for specific metrics to enable automatic detection of those step types. Using basic assumptions about the walk of a cow, the SoftSeparatorTM identifies specific step sequences to recognize where data generated by hind limbs is present and then extracts the data. The results of the SoftSeparatorTM algorithm were tested on 200 data files generated at two different farms. In its best incarnation, SoftSeparatorTM successfully extracted 509 instances of hind-limb data out of 629 total instances, while incorrectly extracting only 25 instances. These results show that SoftSeparatorTM effectively recovers much of the lost data.
Identifier	1069
Format	application:pdf
Language	en
Collection	UMBC Theses and Dissertations
Rights Statement	This item may be protected under Title 17 of the U.S. Copyright Law. It is made available by UMBC for non-commercial research and education. For permission to publish or reproduce, please see http://aok.lib.umbc.edu/specoll/repro.php or contact Special Collections at speccoll(at)umbc.edu.

Add tags for SoftSeparator: A Method for Dynamic Separation of the Signatures Generated by Two Cows Walking in Tandem over the Longitudinally-Split Floor of the StepMetrixTM

View as list | View as tag cloud | report abuse

Post a Comment for SoftSeparator: A Method for Dynamic Separation of the Signatures Generated by Two Cows Walking in Tandem over the Longitudinally-Split Floor of the StepMetrixTM

Find results with:

Searching collections:

Home SoftSeparator: A Method for Dynamic Separation of the Signatures Generated by Two Cows Walking in...

SoftSeparator: A Method for Dynamic Separation of the Signatures Generated by Two Cows Walking in Tandem over the Longitudinally-Split Floor of the StepMetrixTM

Description

Tags

Comments