Precision technology in a real farm environment
Welcome to AgRoboNews.com Ryan. Let us help the AgTechNews.com and AgroboNews.com audience which includes prospective buyers & growers to learn more about your Mushroom Automation.
Niranjan Minase (AgRoboNews.com): If I am a commercial mushroom farmer today, where exactly am I losing money without realizing it? Please quantify in terms of yield loss, labor inefficiency, or wastage.
Ryan McCartney: Farmers are most painfully dealing with labor shortages and issues. Not only is it a challenging job and environment to harvest mushrooms in a cool, humid, dark climate which makes it difficult to recruit for, we’re also facing foreign worker shortages, especially in Canada and the USA. I spoke with a customer in western Canada last week who shared that his comrades at US farms are even paying signing bonuses or wasting crops because they can’t get enough people to pick them.
Once you do have labor, the labor cost has skyrocketed in recent years. In Canada and US loaded labor costs nearing $30/hour, in Europe 25 euro and in Australia, it’s nearing AUD$50/hour when factoring in insurance, benefits and supervisory time needed for harvesters.
Labor efficiency is challenged by all of the micro tasks a harvester has to do. They need to move trolleys around, load boxes, pick, cut and pack mushrooms. Each of these tasks has a lot of room for automation and labor efficiency.
Yield is a big one. Currently growing happens with tribal knowledge and people’s capacity to manage growing conditions through spot checking. With vision systems, AI and machine learning, we can track every single mushroom (millions per room per cycle) from pin to final product, predicting issues, recommending watering or environmental issues that will affect yield, compost optimizations, and exactly which mushrooms to harvest at which time to maximize overall bed yield and meet sales goals.
Niranjan Minase (AgRoboNews.com): At a very practical level, what specific task in mushroom farming does Mycionics automate? Explain it as you would to a farmer who has never seen robotics before.
Ryan McCartney: We automate mushroom harvesting through picking, packing, pointing, and planning automations. Mushrooms are a complicated crop, they have a unique growth curve at 4% per hour, doubling in size daily, and MUST be picked at the right time or they begin to lose quality (and value). The grow in clumps and are often tightly clustered, different heights. They have to be kept clean and they are extremely delicate to touch, how they’re pulled, and temperature, CO2 and air movement and more.
We started in harvest automation with picking robots. This is difficult to do cheaply, with high quality AND fast enough to manage crops. We can now complete all three – on a modern drawer farm. We have modularized our solution starting with our proprietary vision system, which identifies and digitizes the crop. This supports our module arms and end effectors which can be used to pack mushrooms from conveyor belts.
Our Crop Scout solution uses our vision system to scan mushroom beds, create a digital twin, track and forecast results. It also points at the perfect mushrooms to harvest at the perfect time for manual pickers, improving the overall crop yield performance and removing decision fatigue from the pickers who make thousands of picking decisions when cultivating their bed each day.
Niranjan Minase (AgRoboNews.com): What part of the process still requires human judgment, and what part is fully automated?
Ryan McCartney: The thinning or separation phase is important. The small mushrooms don’t grow uniformly, so managing the distribution of growth on the bed to not overconsume the compost, spread out the growth so they dont bruise neighbouring caps, and set the bed up for successfully robotic harvesting is key. Humans are still the best at this part of the process.
Niranjan Minase (AgRoboNews.com): If I invest in your system, what changes immediately on my farm in the first 30–60 days?
Ryan McCartney: Immediately, you’ll have detailed data and analysis of what’s happening with every mushroom on your farm, synthesized into grow and harvest recommendations. You’ll have way more insight into how to improve yield and manage labor. Something that’s never been possible before.
You’ll realize that you can graze harvest, and move to the right location on mushroom beds at the right time. Yields go up almost immediately because your team, or the robots, are picking better mushrooms at better times to optimize yield and quality.
Niranjan Minase (AgRoboNews.com): Let’s talk numbers. What is the typical ROI a farmer should expect?
Ryan McCartney: On Crop Scout for a drawer farm, the payback is less than 1 year. It’s a no brainer for farmers. Plug and play, immediate results, low capital cost. Harvesting and packing robots require more capex and some standard operating procedure change management. We’re seeing 2-3 year paybacks.
On farm trials have shown that customers are experiencing about a 15% labor efficiency or savings. On harvesting robotics, it depends on the number of arms you employ to keep up with the crop. In general we’re about 75% labor reduction on the harvest platform when employing 4 arms.
In our side-by-side trials on farm where one bed has a CropScout and one bed does not, over 6 months and many cycles, we’re consistenly seeing 5-15% yield increase on beds with CropScout. I’m starting to think the price is too low.
Niranjan Minase (AgRoboNews.com): What is the minimum farm size or scale where your solution starts making economic sense?
Ryan McCartney: There isn’t a great answer for this. Any commercial mushroom farm with dutch aluminum shelving or drawer system in place are qualified to benefit from robotics. Most have more than 10 pinning/growing rooms.
Niranjan Minase (AgRoboNews.com): Why is adoption of automation in mushroom farming still slow, despite clear labor and consistency challenges?
Ryan McCartney: There has been a lot of over-promise for nearly 3 decades about robotics taking over and solving the problem. The Cost+quality+speed equation has been challenging with such a delicate crop in a harsh wet environment. Our advantage is that we actually started on, and grew up at a mushroom farm, picking, packing and working on real mushrooms rather than just a tech lab.
The capital cost is no joke. To properly outfit farms is a material capital investment and sometimes a change management effort to adjust to letting the robots operate in the best way. This takes planning.
Successful robotic development requires running lots of cycles. If the market wants perfection, it will take a long time. The early adopters will be the winners, working through iteration to gain progress (and profit) over waiting for perfection. The sooner more of our robots are installed, the better they will become more quickly.
Niranjan Minase (AgRoboNews.com): From your experience, what goes wrong in the first deployment for farmers? Where do they struggle?
Ryan McCartney: Change management is often the hardest thing to overcome - “I’ve always done it this way” mentality. Usually early technology adopters are willing to change and we’ve had great luck with our clients.
Sometimes operating procedures, like shift schedules, box management, mushroom inventory management etc, need adjustment to optimize for robotic performance and increased yields. For example, a graze harvest over a longer period is more effective than forcing an 8 or 10 hour shift once per day, but this is an operational change that has to be experimented with.
Niranjan Minase (AgRoboNews.com): Which type of farmer should NOT invest in your solution right now?
Ryan McCartney: Trying to build robots that fit into non-standard farm formats is hard and expensive. If you’re on a wooden or tray farm, dont’ invest in robotics, invest in a facility upgrade that is futureproofed for robots to optimize. Don’t try to make a robot fit your footprint, its too expensive and ‘custom’ is a recipe for disaster. Upgrade to a consistent, modern format like the Christiaens Drawer system, or Dutch aluminium with blue belt conveyance.
Niranjan Minase (AgRoboNews.com): Mushrooms are delicate. How does your system handle harvesting without damaging the crop or reducing quality?
Ryan McCartney: This is a key challenge to robotic harvesting. We have spent years and millions of dollars perfecting the gripper. We’ve tried all the suction devices, pneumatics, mechanical stuff. Its so challenging to properly pull the mushroom from the bed without bruising, decapitating and removing compost.
We ended up investing a lot in a custom designed gripper and a lot of software built to specifically mimic the human hand movement of picking a mushroom. This includes gripping in the right location on the cap, applying the right force to push, twist and pull form the bed.
The results in side-by-side comparison of cooled mushrooms days after harvest are outstanding – you'd never know it wasn’t picked by hand. Many robotics companies have gotten 80% of the way here, but the last 20% is the most important – doing it with the right force at the right angle and for the right speed and cost! We’ve cracked the code.
Niranjan Minase (AgRoboNews.com): Mushroom farming is highly dependent on environment control. How do you manage variability?
Ryan McCartney: Steam, chemicals, and mechanical endurance are serious obstacles for electronics at this scale. These robots have to be wash-down ready and be steamable. Our systems are all sealed with high tech gaskets and fixtures and desiccating devices and early warning systems inside. Being “Farm hardened” is a must. It isn’t cheap, but it just works.
Niranjan Minase (AgRoboNews.com): Manual vs automated operations, where do I see the biggest difference—speed, consistency, or yield?
Ryan McCartney: Definitely yield. The machines can store years of cycle data and prescribe exactly how, what, when to pick the right mushrooms. Humans just can’t do this. We're talking today about 5-15% yield increases consistently.
Speed is an interesting topic. Sometimes you have to slow down to speed up! Graze harvesting means picking over longer periods. Robots can work 24 hours per day. On average though, robots will still manage a faster overall speed because they can operate at a consistent maximum AND they aren’t doing other tasks like moving trolleys, looking for tools, or moving boxes.
Niranjan Minase (AgRoboNews.com): Looking ahead, what does a fully autonomous mushroom farm realistically look like in the next 5–7 years?
Ryan McCartney: Today we’re still in a world of collaborative robotics or hybrid solutions. Fully autonomous at speed, quality and scale is still out of reach, but in sight. With AI and machine learning on connected devices like our robots, they’re modularly designed to be installed today and continuously improve as we add software layers.
Over 5-7 years, we’ll get to near autonomous with some human interaction still involved as a contingency. The key is deep partnerships with the infrastructure builders – the whole farm has to work together from climate to shelving and conveyance.
Niranjan Minase (AgRoboNews.com): In that future model, how many people are still required on the farm?
Ryan McCartney: It will be far fewer for sure. We believe we can cut labor by 50-75% now with robotic picking, packing, pointing. It means you’ll end up keeping some of your higher valued staff, which is simpler to supervise and they are often more versatile for complex knowledge and judgement tasks like cleaning and inspection.
Niranjan Minase (AgRoboNews.com): Do you see your technology extending to other crops?
Ryan McCartney: Yes, we do. We already have discussions underway with other horti crops and potentially other sorting and packing applications. Our core vision module and intelligence software can be applied in a variety of industries, but for now we’re deep on mushrooms to exploit our domain experience.
Niranjan Minase (AgRoboNews.com): What is the biggest barrier to adoption—technology, cost, or lack of awareness?
Ryan McCartney: Cost is definitely a barrier. The large capex means that we need to get creative with financing and scaleup planning, including yield increase profit sharing.
Technology has been the issue until now because there wasn’t enough proof it could be delivered cheaper-faster-better. We’re getting through that but we do need more early adopters to work with us on creative terms.
Niranjan Minase (AgRoboNews.com): What kind of information actually convinces a farmer to take a decision?
Ryan McCartney: The industry is very much a “show me” kind of audience. Everyone wants to see it working and think through their own implementation. Videos are very helpful; if you’re going to show something, it’s easy enough to have photos or working videos of it in operation.
Niranjan Minase (AgRoboNews.com): How many farms are currently using your system?
Ryan McCartney: We have machines deployed in North America and Europe, and soon in Australia. We’re regularly seeing yield improvements of 5-15% and labor efficiency improvements of 10-15% with a clear path to get to 75% labor reduction.
Niranjan Minase (AgRoboNews.com): Can you share one failure or lesson from deployment?
Ryan McCartney: One key pivot in our evolution was moving away from the “do everything” autonomous robot to modularizing robots to do specialized tasks. This allows you to produce lower cost machines that are easier to maintain and faster at the tasks they’re assigned. When you try to do everything, you limit out on opportunities too quickly.
Niranjan Minase (AgRoboNews.com): What data or information do you need from a potential buyer?
Ryan McCartney: Usually we’re qualifying for technology and financial readiness:
• What infrastructure do you have or are you planning (shelves, drawer, etc)?
• What operating shifts do you run? How do your staff manage harvest today?
• Have you successfully invested in other technologies that disrupted standard procedures?
• What is your level of willingness to iterate vs. expect perfection? This is a land-and-expand relationship.
Niranjan Minase (AgRoboNews.com): What are the red flags?
Ryan McCartney: When a farmer wants a system to overly conform to their unique circumstances, it becomes challenging to deliver a universal product. As an industry, we have to adjust to the capabilities of the robotics, not try to make them do tasks that sacrifice what they’re good at.
Niranjan Minase (AgRoboNews.com): Are you seeing labor shortages becoming a structural problem globally?
Ryan McCartney: Yes. Between government regulations and a general lack of appetite for physically demanding work in an unfavorable environment, it is a structural issue.
Niranjan Minase (AgRoboNews.com): Which regions are adopting fastest?
Ryan McCartney: Canada has been a leader in technology acceptance and development. Europe also has a very compelling business case to switch.
Niranjan Minase (AgRoboNews.com): If I delay adopting automation by 2–3 years, what risk am I taking?
Ryan McCartney: Delayed adoption means you miss the margin gains you can achieve post payback. The sooner you start, the sooner you benefit. You know the financing environment now; this is the devil you know vs the devil you don’t.
The other risk is the “rising tide lifts all boats” issue. The longer adoption takes, the slower the benefits will be. More cycles means faster iterations of smarter machines.
Niranjan Minase (AgRoboNews.com): For a serious farmer evaluating automation today, what is the first step you recommend?
Ryan McCartney: I certainly recommend walking before you run which means staged deployments. Start with a room or two to collect data to justify expanding. We always caution that you have to be planning for success though. We both have to invest in piloting with the intent to iterate and scale as quickly as is feasible.
