Behavioural Monitoring Collars
Collars and yokes have been used for centuries on bovines around the world. Where animals are used for draft this is still the case.
A traditional use of collars has been as a means of locating animals in hill pastures by attaching bells a system that is still used even in technically advanced countries such as Switzerland.
About 30 years ago collars were introduced to carry the original bulky RFID tags and could also be used with mercury switches to detect movement and thus the gross change in movements associated with oestrus. Detecting oestrus has been a key requirement on dairy farms ever since the introduction of Artificial Insemination placed the onus on detecting oestrus (or heat) on the farm staff.
With the invention of low cost wireless tri-axial accelerometers for games consoles and vehicle control a new technical revolution could be envisaged that can encompass far more than detecting of gross changes of behaviour. This revolution is ongoing and eCow is central to the development of new algorithms to detect various health parameters. The engineering principles for cow monitoring with collars were demonstrated in papers presented by Toby Mottram at the ASABE meeting in 2008 at Providence Rhode Island.
The Wireless Sensor Networks_Modelling the Attenuation of Radio Signals by Bovines demonstrated mathematically why the best location of the antenna to be 4-7 m above ground level in a position where the cows head is turned away from the antenna. A further Wireless Sensor Networks for Beef and Dairy Herd Management2 showed that the position of the antenna on the cow was critical to the successful transmission of radio signal from cows particularly those in crowded pens and passageways.
The eCollar utilises this principal by mounting its antenna and monitoring electronics high on the neck in the concavity formed by the neck muscles. It is held in place by a weighted buckle that does not need to be held tight and adjusts naturally to the changes in size of the cow’s neck. If a collar does not fit well a cow can go to great lengths to destroy it as can be seen in this video of a cow finding a sharp piece of barbed wire on a post and systematically a collar.
The analysis principle we have adopted at eCow is to convert the three axes of accelerations expressed in g into posture data, changes of posture and to interpret the posture into behaviour. We feel that this is better than primitive but powerful mathematical models (eg Hidden Markovian) with no understanding of what is happening to the data.
Look at this video of a cow
By analysis of changes of posture we can predict calving by mapping the restlessness of the cow in the last six hours before calving.
See the Lameness page to see how we can use collars to map mobility scores.
to download any of the mentioned papers please register to view them in the library.