Application of CUSUM charts to detect lameness in a milking robot

Matti Pastell, Henrik Madsen

    Research output: Contribution to journalJournal articleResearchpeer-review

    Abstract

    In the year 2006 about 4000 farms worldwide used over 6000 milking robots. With increased automation the time that the cattle keeper uses for monitoring animals has decreased. This has created a need for automatic health monitoring systems. Lameness is a crucial welfare and economic issue in modern dairy husbandry. It causes problems especially in loose housing of cattle. This could be greatly reduced by early identification and treatment. A four-balance system for automatically measuring the load on each leg of a cow during milking in a milking robot has been developed. It has been previously shown that the weight distribution between limbs changes when cow get lame. In this paper we suggest CUSUM charts to automatically detect lameness based on the measurements. CUSUM charts are statistical based control charts and are well suited for checking a measuring system in operation for any departure from some target or specified values. The target values for detecting lameness were calculated from the cow’s own historical data so that each animal had an individual chart. The method enables objective monitoring of the changes in leg health, which is valuable information in veterinary research because it provides means for assessing the severity and impact of different causes of lameness and also evaluating the effect of treatment and medication. So far no objective method for calculating these measures has been available and the methodology presented in this paper seems very promising for the task.
    Original languageEnglish
    JournalExpert Systems with Applications
    Volume35
    Issue number4
    Pages (from-to)2032-2040
    ISSN0957-4174
    DOIs
    Publication statusPublished - 2008

    Fingerprint Dive into the research topics of 'Application of CUSUM charts to detect lameness in a milking robot'. Together they form a unique fingerprint.

    Cite this