Weeds harm plants.  That was our company Tertill Corporation’s mantra, relentlessly repeated to potential customers.  And it’s true; weeds can and do rob crop plants of nutrients, moisture, and sunlight.  This leads to lower yields and poorer quality produce in weedy gardens.  Our robot solved that problem.  Tertill lived in home gardens and prevented weeds from taking hold. 

However, even a weed-choked garden produces some crops—I know this from experience.  

We loved and believed in our product, and it truly worked.  From the time I began using our first prototype, I never again had weeds in my garden.  Even so, Tertill struggled to find its place in the market.   

The problems weeds cause are slow to develop (“I’ll weed tomorrow, for sure”) and difficult to measure.  Those facts made Tertill a hard sell.  We needed customers to imagine the greater bounty they would enjoy if they owned Tertill versus the unquantified bounty they were already getting without one.  But, once I started gardening, I discovered that there’s another garden menace whose effects are immediate, obvious, and maddening.

It was my second or third season as a gardener and I eagerly watched as the tomatoes (in my weed-free garden) matured and ripened.  But often, when I came to gather it, I found that the fruit I’d planned to pick that day had mysteriously vanished.  Slowly I began to notice mangled, almost-ripe tomatoes scattered under the other plants.  A groundhog, it turned out, was beating me to my harvest.

And it wasn’t as though my resident pest had parsimoniously consumed only what he needed for sustenance.  That I would have indulged.  No, his ignoble practice was to take small bites out of numerous tomatoes and knock unripe ones to the ground.  Infuriating!  How dare that lazy bounder deny me the rewards of my hard work!  

My reaction to the groundhog’s predation, I noticed, was much more visceral than my response to the weeds that had plagued my first-season garden.  That suggested to me that our company might be missing an opportunity.  Maybe customers would more eagerly embrace a product that prevented the loss of crops they could see as opposed to future ones they imagined.

Method

How to accomplish this?  Trouble with pests is as old as agriculture and many products already on the market purport to eliminate the problem.  One solution I was familiar with was the ultrasonic animal repeller.  

Such devices often use flashing lights and an ultrasonic horn (that only pests and children can hear) to frighten away the uninvited guest.  Unfortunately, animals invariably become acclimated to the ruckus and learn to ignore it.  But as a test I bought and installed an ultrasonic repeller in my garden.  One day I noticed that a pair of bunnies had entered the garden and set off the alarm.  But oblivious to all the flashing and ultrasonic beeping, they calmly selected their appetizers.  

An effective deterrent needed to be more emphatic, I thought, and a robot seemed like just the thing.  Rather than only making noise, the robot could locate and rush toward the pest.  The pest would then fear that it was in the sights of a predator and, hopefully, execute a rapid retreat.

We might build a standalone robot; one whose sole purpose was to expel pests.  But a useful synergy seemed possible—the pest scaring feature appeared to be fully compatible with Tertill’s weeding mission.  I’d never seen a pest remain in the garden when Tertill was in motion and weeding.  But the robot spent most of its time charging in the sun (at night it slept, waiting for the sun to rise).  The pest-scaring version of Tertill would need to be ready to awaken and give chase day or night.  That required only a little more hardware and a few  changes to Tertill’s program.  

The scaring version of Tertill already had a built-in mechanism to mitigate the problem that pests might become acclimated to it—its weed whacker.  If the pest refused to move when Tertill approached, the robot could turn on its weed whacker and pummel the pest’s toes.  That aggressive attitude suggested a name for the new robot, Tertilnator.  Tertilnator would make the garden safe for tomatoes.

Detecting pests appeared to be mostly a solved problem.  A cheap sensor, manufactured by the millions, makes it possible.  It’s called a passive pyroelectric infrared sensor and it’s the secret sauce in most motion detecting devices.  For instance, such devices can turn on the outside lights at night if someone walks up the driveway.  A pyroelectric detector is sensitive to radiation with a wavelength of around 10 microns.  That’s the peak of the blackbody radiation emitted by people and animals because of their body heat.

I modified an existing Tertill to create a Tertilnator prototype.  I added pyroelectric sensors pointing in four directions around the robot and a pair of LEDs (to simulate predator eyes) pointing forward.  Depending on from which direction the pest approached, different sensors would trip.  That would wake up Tertilnator and give it the direction of the pest.  The robot would first turn on its “eyes,” spin until it pointed toward the pest, and then drive forward, beeping all the while.

Testing

Terminator successfully chased away pests.

The prototypes I built yielded satisfying results.  Squirrels, active mostly during daylight hours, were always obliging test subjects.   To encourage them, I put a couple of peanuts on top of the robot.  A squirrel would approach, the robot would move, and the squirrel would run away.  

I set up a motion activated infrared camera to monitor the robot and garden day and night.  Over the course of a week or two, the camera captured an opossum coming into the garden and being scared away.  A pair of bunnies made an appearance and rapid disappearance.  And Tertilnator also succeeded in chasing away a groundhog.

Tertilnator seemed quite promising, but a good bit of further development would be needed to transform it into a reliable product.  It turned out that squirrels did become acclimated to the robot after a time.  They might have continued to fear the robot if it had whacked their toes as I’d intended but unfortunately, I hadn’t implemented a working whacker on my prototype.

The robot moved too slowly, but that could be corrected by using different motors and gearboxes. Another problem was that, during daylight hours, the pyroelectric sensors sometimes tripped when they shouldn’t have—fooled by the shadows of leaves fluttering in the sunlight. I was working on a filter and some code to mitigate the issue.

Densely planted gardens were likely to reduce Tertilnator’s efficacy.  Too many plants between robot and pest would prevent the robot from detecting the pest.  Also, maneuvering the robot quickly through the plants might be challenging.  I thought about this a lot and came up with several strategies to make the system work better in dense gardens.  One was to place some number of pest-detecting, solar powered sensor modules around the garden.  Any sensor that detected a pest could then call the robot to its location so Tertilnator could chase the pest away.  

Maybe Next Time

My teammates liked Tertilnator and the idea of a sentry robot always on guard in the garden.  But they brought up several objections, some that I agreed with.  First, we had no expertise in chasing pests.   Different sorts of pests inhabit different regions of the country.  Learning to deal with them all effectively, might take lots of testing and lots of time.  Another difficult to quantify problem was the fact that inevitably, the robot would occasionally fail to prevent a pest from having a meal of fresh garden produce.  How effective did the robot have to be to satisfy customers?  We had no way to quantify this.  And there were other objections as well.

Developing Tertilnator into a viable product would take considerable resources and time.  These would have to be taken away from our main product, Tertill.  Ultimately, we couldn’t convince ourselves to accept that risky leap.  

But I’d still like to have a little robot that can scare pests away from my garden.  Maybe I can convince someone else to build it.

Note that this post’s featured image is an actual photo of my (weed-free) garden showing some of the tomatoes I collected after the groundhog mangled them.  But the groundhog itself is photoshopped in from an image taken by D. Gordon E. Robertson and used under CC-SA-3.0.