Marwell Zoo heats antelope enclosure using artificial intelligence

Marwell Wildlife is using artificial intelligence to help heat an animal enclosure in a pioneering experiment aimed to reduce the park’s carbon footprint and reduce electricity costs.

The initiative is a partnership between the zoo and IBM and will test whether technology in a ‘smart’ heater can effectively and efficiently control heating for animals more used to warmer climes.

The technology tested in the zoo’s nyala house uses an algorithm which relies on data from infrared sensors to detect if the antelope, which are native to Southern Africa, are inside.

If the nyala are inside, and the temperature is too low, the heating is automatically turned on.

So far the algorithm is making the right decision about turning on the heating for the nyala antelope 95 per cent of the time.

Duncan East, head of sustainability at Marwell Zoo, said he hopes the experiment will save 30-40 per cent on their heating bills and if successful will be rolled out across animal houses in the zoo.

He said: “We’re really excited to be working with IBM on this novel solution to heating control for large animal housing.

“Typically in zoos large-bodied animals are kept in houses where at least one door is left open at all times so the animals have access to their paddocks. This makes it very difficult, and expensive, to heat the whole house.

He continued: “Instead these animals are kept warm by infrared heat lamps which work by only heating the animals in front of the lamp. A number of lamps around the house provide good coverage but it can also result in heating areas with no animals if they have gone outside or are all together in a different part of the house.

“Up until now it has not been possible to control these heat lamps by sensors so they are left on all night during the autumn and spring when nights are cold and 24 hours a day during winter.

“Current passive infrared movement sensors, as you might use on an outside light, don’t work as the heater will go off if the animal stops moving, for example if they go to sleep.

“This new type of active sensor detects the presence of the animal because it is warmer than the surroundings and keeps the heat on as long as there is a warm body present, regardless of movement.

“It means the organisation will have more money spare to spend on conservation activities or on further developments around the park.”

IBM’s Professor Andy Stanford-Clark, chief technology officer for UK and Ireland, added: “The algorithm we are using in the IBM Cloud enabled us to train the system using lots of examples where we knew there were animals present and no animals present.

“This training has enabled us to create a system that can process the infra-red sensor data with high accuracy, and very few false negatives.

“There is a lot of interest from other zoos, waiting to see how effective this is, and what the potential savings are.”