Disclosure: AFN’s parent company, AgFunder, is an investor in Tevel.
Founded in 2016 by computer engineer Yaniv Maor, the Israeli ag robotics startup set out to solve one of the most pressing problems in modern fruit farming: the diminishing availability of manual labor.
According to the UN Food and Agriculture Organization, the relative number of orchard workers may have halved over the past two decades; however, fruit production has doubled over the same period, in line with rising demand.
Tevel’s answer was to design a system of interconnected drones — called Flying Autonomous Robots, or FAR for short — tethered to a ground unit, which can harvest tree-grown fruits and carry out other tasks, such as pruning, trimming, and thinning, which would typically require human labor.
In the latter half of last year, the Israeli company conducted its first commercial field trials in an Italian apple orchard, with particularly promising results. But the Tevel team has been on quite a journey to get to where it is today – read on to hear more from CEO Maor (YM) below.
AFN: What’s your background, and how did that lead you to launching Tevel?
YM: I’ve been managing and running R&D projects in different companies for 20 years. I graduated from Technion, where I studied computer engineering and system engineering. My connection to ag started 10 years ago after watching a TV documentary about the labor crisis in Israel. [The show asked] 20 healthy, young people to pick fruit – and after half a day, they’d all left. I was shocked that this situation was so bad, so I said: “OK, let’s solve it.”
I went to an orchard to try to understand why no robotics solutions [seemed to work] for picking fruit. Processing power and algorithms were just not strong enough back then; the complexity was just too much.
I waited for the tech to become available, about six years later. My first thought was: let’s build a ground unit with a robotic arm, kind of straightforward. But the calculations brought me to a very expensive and very limited solution, because orchards are tall, trees are thick, and they need more flexible solution. So that’s when I began to think about flying robots, wrote the patent, and started the journey of this company.
From that point it became ‘us,’ not just ‘me.’ I brought in the first team members that I knew from previous projects and companies. [Then we] brought in technical people and agronomists to join the company.
In those very early days, we had a tragedy. One of the co-founders, Oz [Desheh], passed away from cancer; that was a very challenging time for us. His brother Eyal stepped in, became our financial lead [now chairman of the board], and he has been with us ever since.
Other than that, it was like with every startup: you have challenges, you do trials, you improve again and again, every year [your product] becomes faster and faster and more stable; and in the last season, we became commercial – and it took us four years to get to that point.
AFN: How does Tevel’s technology work?
YM: Our technology is multidisciplinary, including AI, machine vision, and trajectory planning and execution. We also have physical things like mechanical engineering to design the unique arms and unique robots we have – it’s not something we can buy off the shelf – and also electrical engineering in the power system. And a lot of software everywhere: in the GPU, in the connection to the cloud, and in the web app that manages the whole harvesting process.
In the last [fruit] season that ended in 2021, everything worked very nicely together. The software and the hardware worked non-stop for weeks. From our point of view, it was a success because we didn’t have any physical problems, no breakdowns.
AFN: What about your business model?
YM: Right now we’re [offering FAR] as a service – so our team, with the equipment, picks fruit for the farmers. This is what happened in [FAR’s field trial] in 2021, and we will continue that in 2022.
However, we are starting collaborations, as we see that as the best way to market. We are integrating our tech into local partners, which could be a vendor that [offers ground-based] machinery [or] service providers that can operate the equipment. We see a nice amount of demand from many parties to do that, and that’s the plan: not only harvesting [trials] on a larger scale, but also the start of collaborations. We’ve already started [collaborations] with two companies in the US and Europe.
AFN: How have the trials in Italy been going?
YM: Rivoira is an early adopter, integrating a lot of technologies into their operations – and this is why we choose them, and also why they chose us. It’s a synergy.
The aim of this pilot was quality. We wanted to show to the grower and to the world that [our robots] can pick as well as human pickers.
The scale wasn’t large, but we worked there non-stop for five weeks and we picked fruit. Every evening, there were two bins – one from our side, and one from the manual pickers. We took them to the sorting house, and every morning, we received two reports; one for the manual, and one for the robotic.
Our results were very good. From our side you saw that, with every software upgrade, [our system] was getting better and better, and by the middle of the pilot we became even better than the human pickers. There was a similar level of bruising, but when we analyzed this, we saw it had nothing to do with the picking itself: there was something in the ground unit [where the robots place the harvested fruit] that needed to be covered with soft material. Fruit was rolling into the bin and the bruising was happening there. Once we [fixed] that, performance was better. This was very important for the grower; they liked that the bruising was lower than with manual pickers.
The second big achievement was grading. When picking fruit selectively [FAR] was significantly better than human pickers, and also more consistent; there was more than 10% difference in grading. That’s a lot of money for the growers.
A simple explanation for this is that, for human pickers, it’s very hard work: they start in the morning and finish late in the evening, and do the same task over and over again. People, after a few hours, are getting tired, not paying a lot of attention to quality. But a robot is a robot: you define the threshold, and it works to that. And when it comes to consistency, it’s very dependent on the picker themselves – if one is paying more attention to quality and the other is less, every day you’ll be getting different results from the human pickers.
An additional achievement was [FAR’s] ready availability. We integrated this new product and didn’t face any mechanical problems, made software updates where necessary, and just worked non-stop. And this gave us the confidence to ramp up production right now and to boost market penetration.
These achievements are all thanks to the team. They are real professionals, dedicated, motivated, and hard workers, that are not afraid of challenges.
AFN: What are your priorities for the year ahead?
YM: First of all, larger scale; and also not picking only apples, but also, in the coming season, peaches, nectarines, and plums, with the same customer.
We have already picked these, but not commercially – so it’s our first commercial pilot on stone fruit. And not only for five weeks [as with the first trial], but two whole seasons.
This year we have about 20 robots. Next year we plan to have hundreds of units in operation.
With funding, we’ve raised over $30 million overall, and this year in Q1, we will start a new investment round. A few strategics want to join, and there’s also a nice amount of demand from VCs. We’re looking for investors who can bring value for the company in its next phase.