Pollination robots are being deployed in greenhouses, but we should think twice before sending them into the field
Imagine that you are wandering through an apple orchard. The sun is shining through the branches and the air holds the scent of delicate blossoms, offering an early nectar and pollen buffet for honey bees and native bees alike. The trees are humming with activity, and you look up to see what visitors you might have. But what you spy flitting between flowers are not bees at all — they are miniature, autonomous robots zipping to and fro, pollinating flowers as they go.
A future with robots pollinating our crops sounds like something straight out of a sci-fi movie, but the concept is not as far-fetched as it might seem. Researchers and engineers from around the world, including the U.S., Japan, Israel, and the Netherlands, have been working on building pollination robots for almost a decade. With major breakthroughs in the last three years, the technology is not far off from regular commercial use.
Not every pollination robot looks like a miniature, buzzing bee, though — some weigh 3.5 kg (7.7 lbs) and fly with helicopter blades while spewing out soap bubbles containing pollen, others are as light as a bee with flexible wings that really flap, and some don’t fly at all, but roll on the ground and pollinate flowers by blasting them with pulses of air.
Arugga AI Farming is a robotic pollination company that has opted for the latter approach. The company is based in Israel, and is the first to commercialize robotic pollination for greenhouse tomatoes. They have developed a robot, named Polly, which is designed to scoot up and down rows of tomato plants while using its on-board cameras to locate and blitz flowers with calibrated streams of air. Tomatoes can self-pollinate, so this agitation is all that is needed to set fruit.
“In greenhouses of tomatoes and other crops, there is the issue of pollination, typically performed using commercially produced bumble bee hives,” says Iddo Geltner, the CEO of Arugga, in a presentation at the International Startup Showcase conference. “They [bumble bees] forage for pollen when they want to, not when it is optimal for the plant, they spread viruses in the greenhouse, and they are also banned in Australia and other countries where pollination is still performed manually.”
Strict biosecurity regulations prevent bumble bee hives from being imported to Australia, a country without any native bumble bee species. Greenhouse personnel are instead employed to agitate the plants using a “pollination wand” — a handheld stick with a vibrating end — three times a week during bloom. Polly currently costs $10,000 each and can handle just under an acre of greenhouse space, but the cost is expected to go down as the company scales up production. Plans are in place for future versions of Polly to include additional functions, like automatic pruning, disease detection, and precise agrochemical applications, which would add value to the investment.
Even in countries where bumble bee colonies are commercially available, like in Canada and the U.S., robots like Polly could be advantageous. In North America, two bumble bee species — Bombus occidentalis and B. impatiens, whose native ranges are in the West and East, respectively — are commercially available. But in the late 1990s, B. occidentalis colonies were suddenly scarce, and an exception was made to transport B. impatiens from the East to the West, provided that appropriate safeguards were in place to prevent escape from the greenhouse.
Of course, some bees did escape, and there are now wild populations of B. impatiens in western provinces and states as a result. The exact impact of introduced B. impatiens on native species is not known, but relying on robots like Polly could eliminate the risk of introducing foreign species in addition to relieving human labor where manual pollination is the norm. And as Geltner mentions, having actual bees that fly from plant to plant can spread disease between tomato plants, like tobacco mosaic virus, tomato apical stunt viroid, and tomato brown rugose fruit virus, among others, all of which hinder plant productivity or marketability of the fruits.
Biobest, one of the major commercial bumble bee suppliers, is actually one of Arugga’s investors, despite robotic pollination technology being in direct competition with Biobest’s product. In an interview with Mike Cherney from the Wall Street Journal, Karel Bolkmans, a Biobest representative, states that Biobest is in the business of pollination, not bumble bees. “If a robot is better than bumble bees, then robots it will be,” he says.
But Polly isn’t the only robotic pollinator on wheels. BrambleBee, developed at West Virginia University, is another robot using a similar concept for pollinating blackberries, which can also self-pollinate. BrambleBee first uses lasers to generate a map of the greenhouse, then drives up and down the rows of plants, logging flower locations in its memory. It then makes a second pass, stopping for a more detailed inspection at each flower and using a pollinating arm with an agitator on the end to pollinate.
The BrambleBee prototypes cannot yet recognize individual flowers (the researchers used QR codes as a proxy for flowers in their preliminary tests), but, given the success of Polly and the advancement of deep learning technology for automated image recognition, this challenge is surmountable. BrambleBee also remembers which flowers have already been pollinated, and doesn’t waste time agitating them again when new flowers call for another round of pollination.
Large, rolling robots were the first to be commercialized in part because they require far less technology development to get off the ground. But they aren’t suitable for pollinating trees, and can’t yet transfer pollen from one flower to another, which is necessary for crops that don’t self-pollinate. For that, flying robots are needed.
Although more versatile, flying robots have two major problems that limit their success: Lighter batteries are less efficient, in terms of their power-to-mass ratio, so it is difficult to manufacture insect-sized flying robots that have sufficient battery life for practical use. And because they are so small, precise movement through the air is difficult to achieve with human controllers, let alone autonomously. But one of these challenges has been overcome by researchers at the Wyss Institute for Biologically Inspired Engineering at Harvard, Massachusetts.
Wyss has been leading the field of insect-scale robotics (defined as weighing < 0.5 g with a wingspan of < 5 cm) since 2013, when researchers there first unveiled the RoboBee, a flying robot weighing just 80 mg — about the mass of a bee. As they published in the journal Science,1 that robot did not have an onboard power supply and had to be tethered to a power source, which obviously limits its practical utility. But in 2019, Dr. Noah Jafferis and his colleagues at Wyss published the first report of untethered flight of an insect-scale robot, which uses lightweight solar cells weighing about 10 mg to generate power.2
But the prototype with the onboard power source, which the researchers call RoboBee X-wing, had poor flight control. In a press release from Harvard, the researchers describe a situation that sounds more like ….