The cow-whisperer: Towards autonomous management of free-ranging cows

TitleThe cow-whisperer: Towards autonomous management of free-ranging cows
Publication TypeConference Proceedings
Year of Publication2009
AuthorsCorrell N., Doniec M, Libeau R.L., Rus D., Anderson D.M.
Conference Name2009 IEEE Interantional Conference on Robotics & Automation
Date Published2009
ARIS Log Number226396
Keywordsanimal, free-ranging, GPS, livestock, management
AbstractAnimal management in the 21st century presents many challenges. Free-ranging livestock are classically controlled by herders. Today, wire fences are used in most developed countries, although they are costly to build and maintain. Furthermore, conventional fences do not foster flexible management of foraging animals. Flexibility is the key to ecological management since both plants and animals are spatially and temporally dynamic resources. Besides logging GPS data, our sensor box allows us to play arbitrary aural cues ranging from irritating sounds to gathering commands. In our first experiment, we investigate how to stimulate a cow to move her in a specific direction by playing various sounds in either the left or right ear, however, none of the audio cues evaluated proved effective. Behavioral theory suggests that a cow will avoid novel cues, in this case sounds, by moving away from them. In earlier experiments using a number of different sounds the loudness and frequency of the sounds used did not have any noticeable detrimental affect on cows. It appears that if we want to control cows without excessive stress, we need to better understand individual as well as herd behavior. In a preliminary experiment to evaluate autonomous gathering, the cows were observed to be in the corral at approximately 7:15 AM drinking water. About mid-morning the cows were returned to the middle of the pasture with an ATV. Past research using these same animals would suggest the cows would not have done what was observed next. They returned to the corral a second time the same day. Forty-five minutes after leaving the cows in the middle of the pasture where they were observed to begin foraging. The preprogrammed instruments started administering audio sounds that the cows would have associated with previous manual gathering. However, instead of immediately walking back to the corral once the audio cues began playing, the cows continued walking south. They reached the southwest corner of the pasture where they stood for approximately 20 minutes. We then returned on the ATV and turned the cows north along the west fence line. Thereafter, the cows continued to walk north along the west fence line for about a mile without further human intervention. Returning to the corral, we observed the cows as they returned to the corral this second time on March 3, 2008. The aural gathering cues were administered equally in both ears to all four cows in 30s intervals throughout their walk to the corral. These 30 s on-off intervals appear to correspond to the stop-and-go behavior one can observe in the video. This experiment suggests that preprogrammed audio cuing involving sounds associated with manual gathering may be sufficient to gather animals autonomously. However, efficient gathering requires movement in the proper direction. Therefore, we believe that directional audio cues may be required in order to establish proper animal orientation. Our next step will be to provide directional audio cues to the animals using hard- and software currently under development at MIT. In addition, we plan to investigate how many animals in a herd we would need to instrument in order to control the entire group.