Killing beans with robots

June 27, 2017 6:04 AM

GreenPiThumb: A Raspberry Pi Gardening Bot
posted by a snickering nuthatch (11 comments total)

This is pretty cool, but as an amateur gardener, his "water distribution" problems drove me nuts. His solution to watering the soil was to create holes in a rubber glove with sewing needles and a nail clipper.

There's a whole world of irrigation drip emitters that would make his life a million times easier.
posted by jeremias at 7:01 AM on June 27, 2017 [5 favorites]

Good on this person for documenting their project and their failures and lessons learned, and the time lapses are super fun, but wow I'm tired of the phrase "but I'm a software guy". There's this very frustrating brand of software developer myopia that presumes that everything that isn't "hacking!" is somehow easy. And so they talk about unit testing, and coding standards, and so on instead of talking to someone who's good at growing potted plants.

"The gardening part wasn't supposed to be hard" has a lot of baggage bundled up with it. I grew up nerdy and steeped in tech culture, and internalized early this idea that non-techy disciplines were "easy". It took me a embarrassingly long time to grow out of it, and it's one of the most toxic aspects of tech culture. So there was a lot of inward cringing reading this account.
posted by phooky at 7:42 AM on June 27, 2017 [25 favorites]

I liked it for the pleasure of self-discovery. They could have consulted with gardeners and irrigation experts, but instead they enjoyed gaining appreciation for the inherent difficulties the old-fashioned way, by re-inventing the wheel. Sometimes that's valuable for your own education, and sometimes it allows innovation that wouldn't happen if you already know the best-practice solution.
posted by TreeRooster at 8:29 AM on June 27, 2017 [2 favorites]

At this point my partner and I are using a direct-feed drip-tape system for our 8-row 35' wide garden. Pressure became something of an issue with the usual single-feed timing systems, so now we're just relying on our own habits to make sure the watering is done every day.

At some point I may look into the OpenSprinkler system if I can beat the pressure issue, but I don't see that need quite yet. I do like the moisture sensor tho, I need to figure out how to cover a large area with sensors with selective monitoring. Hmm...
posted by endotoxin at 9:24 AM on June 27, 2017

I grew up nerdy and steeped in tech culture, and internalized early this idea that non-techy disciplines were "easy".

FWIW, I took that to be part of the author's self-deprecating style, another "look how I screwed it up now".
posted by suetanvil at 9:29 AM on June 27, 2017 [7 favorites]

I really liked this post - he gives a really thorough walk thru of why he did what he did, how to do what he did, and what kind of results he came up with. I'm thinking of getting a raspberry pi and would do something a lot like this with it, so a complete example is really useful to me, a hobbyist. I also appreciate the thoroughness and professionalism that he's aiming for.

He explicitly states "my main motivation with GreenPiThumb was to learn new technologies." He goes on to state that "my friend Jeet had just started learning to program, so I asked if he’d be interested in collaborating with me on GreenPiThumb. It seemed like a good opportunity for him to learn about healthy software engineering practices like code reviews, unit tests, and continuous integration."

Ok, so he does say "gardening didn’t seem that hard." But he's talking about growing a single plant, one he specifically selected to be easy to experiment with: "Green beans, in particular, are frequently described as a hardy plant that requires only basic gardening skills to grow." This is a plant we give to fourth graders and we expect them to make it grow. So really, it's not that hard!

The guy is having some fun, talking about it, and including his friends. He acknowledges other projects, and thanks the people who helped. In this post, I see a good guy doing a fun thing, being open about it, and helping his friend along the way.
posted by everythings_interrelated at 10:09 AM on June 27, 2017 [4 favorites]

I particularly like how he bought an enclosure for the SBC but has all the instrumentation on a protoboard with loose wires all over the place such as would stand up to a determined housecat for maybe all of 20msec.

Also, everythings_interrelated or anyone else reading this and thinking "I need a raspberry pi", please see Why You Should Stop Using a Raspberry Pi for Everything. This project is definitely a fine use-case and notably falls under the first listed exclusion in the article, but in general Raspberry Pis are expensive, underpowered (in terms of computing), overpowered (in terms of standby amperage), closed-source, non-realtime, and generally a poor fit for pin-level IO.
posted by 7segment at 11:15 AM on June 27, 2017 [4 favorites]

The "Low-power always-on servers." use case matches my use of rapberry pi's hanging off the back of tv's in my house, running kodi.
posted by mikelieman at 12:47 PM on June 27, 2017

FWIW, I took that to be part of the author's self-deprecating style, another "look how I screwed it up now".

Yeah that the gardening part turns out not to be easy at at all is kinda the upshot of the whole thing.
posted by atoxyl at 7:15 PM on June 27, 2017

Also, everythings_interrelated or anyone else reading this and thinking "I need a raspberry pi", please see Why You Should Stop Using a Raspberry Pi for Everything.

Respectfully, that article is "yeah, maybe." I mean, yes, but goddamn the Pi is an amazingly capable little machine for not a lot of moola. Its commodity-factor makes distributions easy to roll (though damnit, there should be like 4x bespoke images there are). I grew up with cartridge consoles and shitty PCs running Linux: this thing would have been a wet dream when I cut my teeth and for many applications still is. I'm in for about $70 for a candy-coated, wifi-enabled, DAC-empowered server that I can slap SD cards like De Olde Days with Other, Crappier Media.

I mean, cycle failure? Shit, man, my first media would degrade just sitting there.

I tend to agree with some of the author's vaguer points; you can always get a better job for the task but damnit, its a remarkable little machine for the bux.

There's a whole world of irrigation drip emitters that would make his life a million times easier.

A thousand times this. Every problem in the world has already been solved and can usually be purchased better and cheaper than fabricated. This sort of goes hand-in-hand with "I can do this," though. I mean, the whole exercise is blown when you can get a $10 thingus that does the same thing. Hobbies, man. ¯\_(ツ)_/¯

I build things because the world annoys me with it not doing things my way.
posted by Ogre Lawless at 9:30 PM on June 27, 2017 [1 favorite]

...but goddamn the Pi is an amazingly capable little machine for not a lot of moola.

The economics of embedded devices (which is the industry that these hobbyist computers come from) center around the trade-off between spending money up front versus missing out on potential profits. Is it worth it to spend money optimizing my code to work on a cheaper microcontroller? Is it worthwhile to build a custom chip? (This, BTW, is why you will often see multiple hardware revisions of various electronic gadgets. The product is selling well so the manufacturer has invested in more optimization.)

This is (I think) where the author of the makeuseof.com article is coming from. And if I were going to mass-produce a device that flashes an LED each time I get a Twitter mention, I'd totally be using some really cheap board. But I'm only going to make on or two, so it's worth the extra $35 to have a machine that runs Python quickly and can host a complete development environment.

As a hobbyist, my main optimization is for enjoyment and the Pi is really good for that.
posted by suetanvil at 9:51 AM on June 29, 2017