This engineer has developed an award-winning device to help farmers keep their crops healthier while using pesticides less often.
Five years ago, Lewis Collins was wondering how, living on a PhD scholarship, he could fund an idea to automate spore trapping. This year, his company BioScout will install its 130th unit and make nearly $2 million in revenue as it continues to expand across Australia, New Zealand and Canada.
BioScout units are autonomous stations that track airborne diseases, currently mainly in agricultural settings.
Artificial intelligence helps the company’s morphology experts to identify pathogens and particulates unique to each site captured by microscopic images. These can provide a complete picture of disease risk weeks before there are visible signs on the crop.
The system works by drawing air into a box where particulates of interest are aerodynamically separated out of the airstream and trapped on an adhesive film.
Microscopic images from the film are sent back to head office, where morphologists manually assess the images and teach artificial-learning models to automatically detect disease in near real time. The daily results are shared with the customer via custom software.
The innovation won the 2023 Good Design of the Year Award.
The right skills
It’s a long way from when Collins was looking for an honours project as part of his PhD in engineering in 2018.
His supervisor, who had a background in spore trapping, suggested Collins had the right mix of skills – mechatronics, biomedical, agricultural science – to look at automating airborne disease detection.
Collins suspects his idea hadn’t been attempted before because of the complexity of the problem.
“You need to have a biology background, especially in crop pathology and mycology, the robotics knowledge and all the many disciplines within that – electrical, mechanical, etc. – and then also understand the issues facing farmers,” he said. “It’s a very integrated piece of work.”
Collins said there were several challenges in achieving the automation.
“I was inspired to try that technology because at the time the process was to either use laborious genetic testing or even just a manual tally of the spore counts while looking through a microscope. Year 10 biology stuff,” he said.
“Then I was at a hospital looking at automated biopsy machines that cost a couple of million dollars, but they were doing similar work where they were automating the scanning of the biopsy slices of tissues. So I took that idea and wondered, ‘How can we do that for $10,000 and put it on a farm in the middle of nowhere?’
“That was a big challenge. A lot of my PhD was focused on building the algorithms to be able to get the autofocus imaging process happening, and then AI was the last bit I tacked on just as a proof of concept.”
While developments in AI have helped BioScout to offer the right tools at the right time, Collins said a lot of factors combined to get the company to this point.
“The [Internet of Things] revolution, better 4G comms across Australia so we can get further distances, hardware costs have come right down, supply chains have improved, and even rapid iterating 3D printing has helped – our first prototypes took 60 hours on a 3D printer,” he said.
“Without all of those things, and without the AI and without having computers that are cheap enough to put in the box, we wouldn’t have been able to get it going.”
Not-so-old thinking
Despite all this innovation, when the time came to take the product to market it was initially dismissed as an antiquated approach.
“The work in plant pathology has moved to genetics, so when we suggested getting back to the old school ways of looking at the morphology of the spores through physical microscope images, we were told it was an outdated approach,” Collins said. “It took a bit of convincing that you can identify numerous fungal spore species just using the morphology of the spores, especially if you get a big enough sample size.
“You can almost see the evolutionary changes with AI.
“So that was a big piece of education and now those same people that were a bit sceptical are our biggest supporters.”
“We now work with basically every crop pathology group across Australia doing analysis and planning our trials. It’s been a big turnaround: conversations which were ‘genetics is the answer’ are now ‘sometimes it’s just better to use BioScout’s morphology approach because the genetics on a farm can be so mixed up and noisy’.”
Increasing applications
While a vast majority of the now 130 BioScout units are in Australia, there are also a few in New Zealand and two in Canada. And interest is expanding beyond agricultural applications, including working with Melbourne University to look at early pollen detection for thunderstorm asthma.
“The best thing about our systems is that they’re picking up every single particulate across Australia, every day,” Collins said. “We can choose which pathogens we’re focusing on for that customer, but in the background we’re building up this big database of where all these particles are and where they’re going.
“I’d say it’s the world’s largest kind of disease detection network. There’s really nothing else that can detect such a large range of micro particulates at the scale we’re doing.
“We’re looking at these microscopic particles which are between 1 µm to 500 μm across the 4000 km breadth of Australia.”
Collins said BioScout is also expanding to cover more dangerous particulates, such as asbestos, and is working with biosecurity experts to detect exotic diseases before they enter Australia.
He said it has been a “crazy experience” over the past five years, going from a PhD student to head of a company with 16 staff. He is just about to close a round of investment funding that will allow the company to fully enter new markets, including North America.
“We’re also expanding our crop offerings,” Collins said. “Our main focus to date has been in cereals and wine grapes, and now we’re moving into tree crops such as almonds and apples, which have both disease issues as well as pollination issues.”
Until now, farmers had to either be reactive with their disease control – spray when they see insects or fungus, which is too late – or proactive, which can lead to overspraying.
Collins said BioScout has given the farmers the confidence to go from spraying every week to every two or three weeks.