Some dogs have jobs and others are pampered pets, but how can we tell the difference? Smart toys that accurately predict which pooches have the right stuff to become guide dogs may be the answer, saving time and money.
In the US, 60 per cent of dogs that start guide dog training never finish it. The process is arduous, taking two years to complete and costs around $50,000. To try to predict which dogs will succeed, a team at the Georgia Institute of Technology developed a sensor-filled ball and tug toy. They then gave it to a dog training centre to use with a new cohort of 40.
Advertisement
After collecting two years鈥� worth of play data, such as duration of biting, force of bites and frequency of bites, 鈥渨e were able to find patterns that could go on to predict the outcomes with 87.5% accuracy,鈥� says Melody Jackson, one of the team.
Due to the nature of the analysis, it wasn鈥檛 possible to say whether one particular way of biting or playing will make a good guide dog and another won鈥檛. The machine learning algorithms used are notoriously difficult to interpret. Despite this, the team say that using the smart toys to identify the most promising dogs in a similar cohort size could save more than $70,000, with testing only taking a few days.
The toys could also lead to more effective training, says Elaine Cannon, at Irish Guide Dogs, a charity. 鈥淎ny test which can give us an insight into a dog鈥檚 potential is very useful and can help us tailor the training or select the trainer from the outset so they can unlock their potential,鈥� she says.
The team hopes to use the smart toys with police dogs, search and rescue dogs and other working dogs in the future.
Reference: Interactive, Mobile, Wearable and Ubiquitous Technologies,
