Automation in Vertical Farming: Plugging the Gaps

The term ‘automation’ in the vertical farming sector is often greatly exaggerated. Press releases regularly refer to ‘full,’ ‘total,’ or ‘complete’ automation, or similar. When digging beneath the surface, this usually refers to basic operational elements (primarily lights going on or off, nutrient delivery, and modulation of environmental conditions against a baseline). This is not the case for all vertical farming companies, but it is certainly widely prevalent. But what does automation really mean in the context of a vertical farm, and how critical is it?  

Compared to other commodities and fast-moving-consumer-goods (FMCGs), fruit and vegetable production (both outdoor and indoor grown) is primarily low margin and highly dependent on scale to achieve financial sustainability. Broadacre farms and greenhouses typically work to single digit margins which are often only achievable at significant scale, especially for the more ‘staple’ products that go beyond microgreens and herbs. The larger the product and the longer the grow time: the higher the risk and harder the unit economics. Indeed, this could be why hydroponically farmed greenhouse tomatoes or whole head lettuces require tens of thousands of metres of production space before the unit economics allow for sale at a competitive price point and subsequent commercial viability. It is important, after all, that the industry pursues a mass market price point that addresses food sustainability at all levels of the socioeconomic ladder – deviating away from high-end, niche products.

Basic or ‘generation one’ vertical farming systems of the past 5-10 years have shown improvements in output per m3, grow time, product quality, and other similar metrics compared to broadacre farms. Specific light and nutrient diets combined with optimal environmental conditions and consistent light exposure are the main contributors to these improvements. However, these improvements alone can be insufficient, especially considering the increases in energy consumption, often higher rental costs, higher CAPEX, and other factors associated with running a vertical farm – irrespective of location.

Vertical farms function across a broad range of activities and stages, with efficiency being critical for each. These include seed treatment, seeding, germination, tray movement across the facility, harvesting, packing, monitoring, and cleaning. For example, for a 500m2 vertical farm (which is small) that grows microgreens, baby leaf greens, and lettuces, seed treatment and seeding (two of the most critical activities to get right) can alone be incredibly time-consuming, OPEX-intensive activities. These two activities would typically require around two full-time staff and when considering wholesale and distributor price points and resulting revenues, these roles alone have the capability to have a substantial impact on margins. What is worse is that roles and activities like these do not necessarily improve with scale: the need is still there, and, with a larger facility, there comes greater risk for human error.   

Multiple players in the vertical farming sector are making inroads in many of these areas. For example, seeding automation across multiple seed types is becoming less rare, as are some forms of dynamic harvesting. However, farms that grow a broad range of produce types and use different forms of substrate and/or growing methods have the added complication of incorporating these into these automated processes, scripts, and programming. Over the past few years, some players have opted to incorporate autonomous robots in their growing environments, the efficacy and efficiency of which is still yet to be determined, with some suggestions of it being mostly modern theatre.

Similarly, scissor lifts, assisting staff up multiple stories worth of farm racks, still appear to be commonplace in even newer vertical farms, used alongside the beforementioned robots, with the latter fulfilling the simpler task of tray transportation. These can pose significant health and safety risks for all farm operators, are wholly inefficient as they require human labour, and have operational limitations. Moreover, their continued use is a sign of acceptance that direct human interaction is still required at the plant level, in existing solutions, even at larger scale. For broadacre farms, mechanisation at scale is also considered to be more advanced when compared to the average vertical farm.

Our views and experience at Vertical Future are different than most. Our systems offer full end-to-end automation, which we define as 'significant human involvement at only the seed treatment and product packing stages, with the rest of the process requiring monitoring of data points only and limited human involvement.' A fully automated vertical farming system should focus on minimising human involvement at every stage of the production cycle, with exceptions at the very start (seed treatments and priming) and end (packaging and distribution) of the entire process. Leveraging such a system must rely on a centralised control framework, tracking, storing, and monitoring key performance metrics. Benefits of this approach include (a) elimination of unnecessary operational bottlenecks; (b) drastically limiting the need for human interaction with crops; and (c) increasing site health and safety for operational staff, through rejecting common industry practices, such as the use of ladders and scissor lifts.

Like a car production line, we divide the vertical farm into strict operational processes and apply low-energy forms of automation to facilitate a plant’s journey through the farm, from seed treatment to harvest; making use of gravity, water-driven transportation, and using energy only when absolutely required. We believe that human involvement should be kept to an absolute minimum and, by using our smart central software platform, Diana, and a smart approach to sensors, we can enable this.

Some aspects of our system are complex, but over 80% is simply the smart assembly of rigid, functional processes. Across the wider industry, we have found that there is often a need to overengineer basic auxiliary functions for the sake of looking impressive, especially considering the age and maturity of the industry. But there is so much information from other industries already out there that can be readily upgraded, adapted, or replicated to improve the operational efficiency of a vertical farm. This is therefore not a question of reinvention in some cases – but more so of innovation through adaptation across a broad range of functions, supported by a complex data platform that ‘conducts the orchestra.’

What are your views on automation in vertical farming? Continue the conversation with us!