While it's important that dogs know some basic obedience commands at the very least, training them can be a monotonous and frustrating experience. Well, perhaps before too long, we could have computers doing the job for us. They're already being used to teach dogs to sit, at North Carolina State University.
Utilizing technology from a previous project, the system requires dogs to wear special harnesses. Sensors within those garments monitor the dog's posture and movements. Each harness also contains a microcomputer that transmits data from the sensors wirelessly.
In the recent experiments, the system was used to determine when dogs went from a standing to a sitting position. When they did, an automated dispenser emitted a beeping sound, and gave them treats.
To create the system, a team led by professors David Roberts and Alper Bozkurt studied 16 human volunteers and their dogs. In particular, the researchers were attempting to arrive at a trade-off between giving the rewards fast enough to be effective, and yet waiting long enough to make sure that they were rewarding the right behaviour.
In its present form, the system rewards the appropriate behaviour 96 percent of the time – that's still not up there with the human trainers, who manage 100 percent. It is more consistent in dispensing rewards within a given amount of time, however, which is important in its own right.
"This study was a proof of concept, and demonstrates that this approach works," says Bozkurt. "Next steps include teaching dogs to perform specific behaviours on cue, and integrating computer-assisted training and human-directed training for use in various service dog applications."
A paper on the research was recently published in the International Journal of Human-Computer Studies.
Source: North Carolina State University
