UK-based Agri-tech manufacturer Vertically Urban and global tech company Microsoft both believe that CEA not only has the power to play a pivotal role in ensuring food security for future generations but believe they can bring about significant improvements in the efficiency and quality of the process.
Developing HORISS with Microsoft AI
The two companies are working together to develop Vertically Urban's AI-based CEA control platform HORISS. The aim of HORISS is for it to be an intelligent control system that autonomously monitors and reacts to environmental and physical changes within the growing area, ensuring crops are grown to exact specifications, using precisely the resources required.
Andrew Littler, Vertically Urban's CEO, sat down with Jens Hansen, General Manager (EMEA) of Data & AI at Microsoft, to discuss Microsoft's partnership with Vertically Urban and how AI has the power to accelerate the new farming revolution in CEA and secure adequate food supplies for future generations.
AL: Thanks for hosting us here at Microsoft HQ, Jens. It's great to be able to sit down and discuss working together on our plans for the future!
I think it's fair to say that farming wouldn't be the first thing in people's minds when they hear the name Microsoft. Why is this area of interest to a company that, it's fair to say, people would often associate with the workplace?
JH: First of all, let me start by saying how incredibly excited I am about the partnership between Microsoft and Vertically Urban.
To answer your question, fundamentally, Microsoft believes that technology is a powerful force for good and has the potential to create novel solutions to many of humanity's issues. In addition, we think that it's important to ensure those solutions, in turn, lead towards a sustainable future where everyone has access to the benefits and opportunities created by them.
With that in mind, what is a more important use of technology than to ensure humanity has the fundamental resources to thrive both now and in the future?
With both that and the goals of the HORISS platform in mind, I think we're talking about a very relevant topic today, food security.
AL: We see it as a full package. We can offer software (HORISS) and hardware to enable all these farms of the future to become the providers of our food.
JH: I know that the team is super excited to work with your organization on that. So basically, the HORISS software, if I understand correctly, monitors the growth process in the plant and adjusts the environmental parameters, and that is intelligent in a way.
AL: Correct, Jens. Essentially, if you look at the AI technology using sensors, camera vision, and algorithms, what it's effectively doing is monitoring crops' performance. Making sure that they're getting fed the right nutrient dose, the right amount of light, and all the other things that are required to achieve the desired growth.
Microsoft's AI capability will excel in the processing and analysis of the vast amounts of data collected by the HORISS system. The aim is to identify trends and patterns in the data that allows us to not only recreate conditions that result in the most desirable end crop but eventually to automatically spot and remedy issues occurring during the growth cycle, crucially, well before they would be spotted by a human resulting in minimal impact to the final harvest.
The more data HORISS harvests and incorporates into the model, the earlier it'll be able to detect issues and the more effectively it'll be able to remedy them in a timely manner reducing waste of resources or, indeed, failure of the whole crop.
In a nutshell, what we're looking for is an optimization that maximizes crop yield while minimizing resource use, and AI helps that optimization. How can you see the way that Microsoft can help us enable (that) in the future?
JH: Good question. Microsoft is a platform company first and foremost, so we're trying to always be at the forefront in terms of new developments in these fields, including various AI technologies with object recognition and other sorts of subdomains in that space. But I think we also have a responsibility to partner with companies like yours, to bring this to market together and enable scenarios that, in your case, actually are for the better, for the good! That's why I'm really excited to see a partnership like ours come to life.
So far, we've talked about satisfying demand in terms of raw volume, but in most cases, the final crop also needs to taste great for it to be classed as a success. We all enjoy eating tasty food, right?
Machines and computers obviously can't taste a crop directly. That's the domain of humans, so it's vital that we find other ways to ensure that whatever we grow also tastes as desired.
This is a great example of a specific challenge HORISS and Microsoft AI can take on, isn't it?
AL: Indeed, creating a proxy for taste is both the most challenging and interesting part of the process. In other words, can we find a set of growing conditions that guarantee a great-tasting harvest? Furthermore, can we get to a point where we know the specific influence of each of those variables?
Within a HORISS-controlled CEA system, there will be a detailed record of the conditions that each crop experienced over its lifetime. If we flag the outcomes as desirable and undesirable, those datasets can be combined and analyzed by Microsoft's AI. Over time, the system will be able to calculate growing conditions that not only ensure an optimal crop yield but ensure it'll look and taste desirable.
Ultimately only humans can decide if that last part is a success or not.
JH: Absolutely. It's important to remember that while AI is a powerful tool, it is not a panacea for everything, nor is it a total replacement for humans. If anything, it takes on the more repetitive and intensively analytical tasks, freeing up humans to do other, more complex work that AI just isn't suited to.
For instance, if we look at the relatively simple concept of picking an apple from a tree. A human can easily navigate stray branches and leaves, assess the ripeness and adapt to the strength of a stalk. Plucking the bounty from the tree with relative ease and, in many cases, without really looking. AI-controlled robots struggle with this task as there are too many variables that don't fit within a pattern. The apples may be hidden behind a branch or leaf; An apple may look ripe from one direction but not another; Two identical-looking apples may have different stalk strengths, etc.
AL: Good point. I think it's easy to get carried away with what you see in the movies compared to what is actually possible in the real world!
One of the reasons that CEA is particularly suited to AI control is that the regular layout makes sensors much easier to be located, and it removes many complex variables from consideration. This makes it much easier to collect valid data and, therefore, truly useful outputs.
What would you say is the end goal of Microsoft's AI involvement with Vertically Urban, HORRIS, and the wider CEA industry?
JH: We'd love to see a future where all of humanity has a reliable and sufficient food supply that is both tasty and nutritious. All done with less waste and minimal impact on the environment.
AL: Yeah, I think food waste is not talked about enough. I remember recently seeing a statistic that around 19% of all lettuce grown in the UK ends up going to waste, which is crazy.
There's a lot of investment by the big supermarket chains in the CEA sector, especially in vertical farms, to secure supply. It'll be interesting to see how smartly they utilize them to reduce waste too.
They (supermarkets) already use the power of AI to predict shopper behavior and control their procurement and distribution network. Integrating an AI-controlled smart vertical farm into that supply chain would allow produce to be grown only as required, which has a huge potential impact.
JH: Indeed. It'd be great to see less produce going to waste for sure! Our conversation has really been an eye-opener, and I look forward to learning more as time goes on.
For more information: