Machine learning builds a better bed for nyalas
We've produced an innovative application of Internet of Things (IoT) technology in the field... literally, in the field.
Marwell Wildlife has built better beds for their animals with IoT and machine learning. They started with nyalas, deer-like creatures that are part of Marwell Wildlife's education, science, and sustainability goals - conserving species and their habitats, locally and globally.
Using Watson IoT Platform, the park’s keepers have worked with IBM to reduce their carbon footprint and create a cosy environment for their animals. All without compromising habitat quality.
And as you can see from our photo, the nyalas are not camera shy.
Components installed in the animal living quarters at the zoo include thermal imaging sensors and Raspberry Pi hardware and a spot of 3D printing, in conjunction with IBM Cloud, the Watson IoT Platform, and technologies developed by our Hursley technicians: Node-RED and MQTT.
It means that the heating comes on only when the nyalas are present, and stays on while they remain in their sleeping enclosure.
From this prototype, the team trained a neural net classifier based on the gathered data, and improved animal recognition. Extending this "nyala neural net" led on to considering further instrumentation across the park - for the comfort of the animals, the efficiency of the park, and even to enhance the experience of visitors.
With the help of business partner Infinitely Simple, the second generation sensor will incorporate external temperature data to increase savings, and expand the trial to more complex housing with different animal behaviours.
Beyond that, there's further potential for other zoos, theme parks, and even public transport.
The solution detects the presence of animals, but can also be used to identify what queue lines are like at the zoo. Come and see our demo, to see how it can detect you. And if you're doing your impression of a nyala, don't forget to smile, deer.