Virtual Fences for Controlling Cows

1
Virtual Fences for Controlling Cows

• Presenter Serafettin Tasci

2
Problem Definition

• Herding Animals are rotated frequently between
  paddocks to prevent overgrazing of any one
  pasture.
• Goal of the study constrain the location of the
  animal by a moving virtual fence algorithm
• A first step toward automatically controlling the
  location of individual animals as well as the
  herd
• It can also be used to monitor the grazing
  behavior of animals in order to create models
  that will lead to better land and pasture
  utilization

3
Challenges

• Short battery life
• About 2 hours 40 minutes
• Lack of fully water resistant devices especially
  for rainy days
• Many components should be combined into a single
  unit
• Easier water resistance and higher battery life
  possible
• Usage of sound stimuli due to stringent
  requirements
• Electric shock is known to be more effective but
  cruel

4
System Requirements

• A smart collar consisting of a GPS unit, a Zaurus
  PDA, wireless networking, and a sound amplifier
• Zaurus PDA 206MHz Intel StrongArm processor,
  64MB of RAM, with an additional 128MB SD memory
  card
• Embedix Linux with the Qtopia window manager
• Wireless network connection 802.11 compact
  flash card
• GPS Unit eTrex GPS unit is connected to the
  serial port of the Zaurus
• Sound amplifier A small Smokey brand guitar
  amplifier

5
Implementation Issues

• Fence defined as a point on the surface of the
  earth, a safe direction, and a velocity.
• Message Handling Wireless network and Unix pipe
  messages
• Multi-hop ad-hoc networking because the pastures
  are too large for single hop messages to reach
  all the animals.
• But generally, the number of hops required for an
  Alive message to reach the laptop base station is
  very low since cows tend to move together.
• Logging and Time Synchronization All log entries
  are accompanied by a time and date stamp.
• GUI add buttons for triggering sounds on
  specific cows, a map showing current cow
  locations and status, and a status display
  showing whether Alive messages have been received
  recently from each cow.

6
Experiments

• Matlab simulation
• Aimed to test the various virtual fence
  techniques
• Real-world physical experiment
• Done with 10 smart collars on cows at Cobb Hill
  Farms in Vermont
• Targeted at 4 issues
• collecting data to create a grazing model for the
  cows, which is used in the fence control
  algorithm
• collecting connectivity data and information
  propagation data, which is used to determine the
  multi-hop routing method for networking the herd
• collecting stimulus response data for individual
  animals
• collecting response data for the virtual fence

7
Results

• The cows have a wide range of speeds throughout
  the day. They spend a large amount of their time
  moving quite slowly, and the rest of the time at
  higher, but differing, speeds.
• Some cows definitely reacted strongly to a sound
  stimulus, though they often quickly became inured
  to it, and stopped reacting.
• The cows noticed the sounds, but either ignored
  them or did not make the desired association with
  their position.

8
Future Work

• Run experiments with beef cattle on an open range
  in Australia
• Since they are wild, they may be more sensible to
  sound stimuli
• Help of animal behaviorists can be used in
  designing an effective collar system and virtual
  fencing algorithm

9

• Questions??