New artificial intelligence sensor 'smells' disease on farm

Artificial intelligence is being used on-farm for instant disease detection
Artificial intelligence is being used on-farm for instant disease detection

A sensor has been developed that ‘smells’ when a disease is present in livestock housing and crop storage.

Cambridgeshire based biotech firm Roboscientific have developed a device that can immediately identify a disease or virus in an enclosed space.

The company says that this will alert the farmer as soon as the livestock or crop become infected.

Whenever a living organism is infected with disease it gives off a specific smell. The device is able to sample the air for the volatile organic compounds (VOCs) that make up those smells and alert the farmer when it identifies the specific VOC fingerprint of a disease.

The technology holds huge potential, as it is able to cut out the slow process of PCR testing to identify infections in livestock.

Currently, Roboscientific are fine tuning the device for commercial sale on chicken farms but are also developing the same technology to identify Covid-19 in spaces such as classrooms and aeroplanes.


Angela Curtis, sales director for Roboscientific, says that her husband Stan, who formed the company, came upon the idea when he was working in Africa.

“He came across a company that was playing around with VOCs for human health," she said.

"They carried out a trial at an African field hospital to see if they could conduct a TB test that was able to identify sufferers on the same day as they were tested”.

Angela explained that this was a success and they identified 95 percent of sufferers on the day.

“From there we decided that agriculture was the most commercially viable route for this technology because everything is more long-winded and expensive medically."

The gas oxide sensors that were used in the initial African TB trial proved to be inconsistent and had a relatively short lifespan, so in 2015 Roboscientific began developing the reliable, hypersensitive sensors that their device now uses.

How does it work?

The device allows air to pass through it, picking up the VOCs that are in the atmosphere using a carbon pad.

It then analyses those VOCs using sensors that identifies their concentration in parts per billion, and it is the pattern between those different VOC concentrations which identifies whether a disease is present or not.

This will then alert the farmer using text or email and tell them exactly what disease or virus it is.

The device samples the air twice a day for around an hour, which allows an accuracy to be reached which, according to the data the scientists have gathered, is close to 100%.

As the device can only detect the VOC fingerprints given off by active diseases and viruses, it is much less likely to give false results, unlike RNA or DNA tests that can react to residual DNA or RNA that is left behind by dead viruses.

At present the device is mainly being used to identify diseases, particularly in chicken houses, however it is able to identify viruses as well.

Although viruses do not give off VOCs like diseases do, the symptoms they generate in infected organisms create metabolic changes which do give off VOCs.


The device is not yet commercially available as Roboscientific are still rolling out the final trials to prepare it for broiler farms.

Angela explained: “This is nothing to do with the technology itself, it is the data processing, we want to make sure that the software gives a straightforward pass/fail signal to the farmer.”

Angela said that the production model for chicken farms should be completed by spring this year and that they will be costed at around £5,000 a device, with two required in a shed of 50,000 broiler hens.

The company says it is straightforward to add different diseases and viruses for the device to detect as it is simply a matter of identifying their digital fingerprint and including it in the sensors, which opens up huge potential for the technology.

Roboscientific are currently running feasibility tests on dairy cattle to identify para TB, as well as looking at different viral diseases in pigs.

The device is also in more advanced stages of development to be used in vegetable crop stores, particularly for fusarium in onions.

Angela said: “The most important period for monitoring onions is in the first few weeks of storage because if fusarium is in them at that stage the farmer knows that those particular onions need moving on the fastest.”

The device can be used in potato stores as well, however the major difficulty comes when the potatoes are stored in boxes because it identifies that disease is in the store but not which of the many boxes it is in.

In action

David Speller manages some of the broiler farms where Roboscientific have been trialling their devices, he believes that this technology has the potential to increase the effectiveness of preventative treatments.

He said: “The issue at the moment is, how can we pre-empt illness in our birds so that we can treat them before they show symptoms?

"The products we use tend not to be antibiotics but preventative treatments, so by the time they show symptoms it is too late.

"At the moment we are wasting money on drugs that the flock might not need because you have to administer it 5 days before a symptom.”

David also says that the technology is more effective than other sampling methods.

“If I go into the shed and take some boot swab samples or faecal samples, that is not representative of all the birds in that shed.

“However, the air in that shed has done a lot of mixing before it is sampled by the device and so is far more representative of the whole shed.”

David believes that the success of the technology will depend on how the collected data is presented to the farmer.

“It is the software that is crucial, there is a lot of information coming off of the sensor and we need to know what it means.

"At what point does the farmer need to act and at what point is it just a bump in the road?"