Hacker News Reader

Comment by TaylorAlexander | Hacker News Reader

If anyone is interested, I’ve designed a totally open source 3D printable off road robot that is meant as a software development platform for farming robotics. I’m currently working on improving the longevity of the gearboxes and on a camera based vision system using reinforcement learning.

Some videos:

https://youtu.be/LIkX7wniYh0 https://youtu.be/cU_0M1_TvD0 https://youtu.be/DXPmqCd0r04

The whole robot and software is licensed CC0 and BSD respectively. Onshape and github links are on the site I promote in the videos.

Would any of you want to build something like this? What would you use it for? Anything you would want that this one is missing?

View on HN

Have you looked at OpenTorque actuator?

https://hackaday.io/project/159404-opentorque-actuator

Not sure how appropriate the form factor is for you, but maybe you can at least find some useful ideas for your own gearbox design.

Hey that’s killer! I’ll have to dig in to it more, thanks for the link!

I would love to build something like this just for general personal experimentation including a kid. I basically want something that is:

- somewhat large (about this rover size) so it can drive around a typical residential yard mixed in with occasional obstacle

- sturdy enough to carry payloads like 1/2 gallon of water or so, survive a kid sitting on it, etc.

- relatively water resistant. Doesn't need to survive falling into a pool, but operating during light rain and especially dew would be great. Needs a payload housing that would protect hosted electronics from getting wet.

I'm not even sure how much the suspension is needed, I feel as long as the wheels are independently driven and relatively large that would be OK.

Mission #1: chase away squirrels and rabbits

Mission #2: shoot them with a water jet/acorn launcher :-)

Mission #3: real-time video feed and remote control to explore the woods at night

For waterproof electronics, I've had success building with these types of cases, just drill holes for wires going out, and seal them up with Sikaflex. With the transparent ones you can have an OLED display on the inside for status etc. They're usually found in your Home Depot style shop:

https://www.ebay.com/itm/Detachable-Fishing-Lure-Bait-Hooks-...

That looks awesome! Here's a comment I wrote earlier about the topic:

I've been thinking about a suite of robots on the residential level that could not only mow the grass and edge it, but also shovel snow, and perhaps eventually prune plants and pick the harvest. If a base, plus a suite of attachments, could be affordable at less than $5k retail, I would buy it.

Awesome! That price might be hard to achieve any time soon. Mostly because the engineering team will be very expensive, so the price will have to be many multiples of the unit cost for some time. Such a thing might be possible at a $20k price in five to ten years, and a service company could offer that work on contract to many different households.

As robotics matures as a field, the engineering costs to accomplish this will go down, as will per unit costs. Maybe in 20 years time such a multi bot could be available. It’s hard to predict the future and probably of questionable merit. That said I’ve become much more pessimistic about timelines for robotics as I’ve continued to work in the industry.

My hope though is that open source robotics can keep pace with the industry as time goes on, just as open source operating systems have kept pace with their closed source counterparts. That way people will have options when the technology does become feasible.

Why, in particular have you become more pessimistic about the timelines for robotics? I've also been headed this way, but it's as a complete outsider of the field simply extrapolating directionality comparing promises to deliverables with the former exponentially outpacing the seemingly linear, at best, progress of the latter. Interestingly the same is also true of many other fields from genetics, to AI, to molecular nanotechnology. The latter being a radical example of something going from 'the next big thing to revolutionize society' to all but disappearing in a period of < 10 years.

Thanks for the insightful comment. If you ever want to hop on a call and talk about micro-farming and robotics, I'm all for it.

Also, after some thinking... what's the current price of just the parts?

About $1500 including batteries. $400 of that is motor controllers, which could be reduced to $130 using Odrive instead of VESC for motor controllers. This is for the remote control version. If you want it self driving, consider the cost of a beefy mobile computer and sensors. I am currently working on a reinforcement learning system that will use cameras and the $1300 NVIDIA Jetson Xavier computer for autonomy.

What about something like this, but without the person controls? https://bealoving.com/products/electric-scooter

Or this? https://innovgoods.com/products/snow-plow-robot

Ah, this seems to be a scam, according to the actual manufacturer: https://www.atr-orbiter.net/

Thanks again!

Stop growing grass, plant bushes, and clover. Don't use gasoline lawnmowers. Lawns are very silly things.

If you don't care about energy costs, you could put a hotwater system under your driveway, I've seen a few of these now.

I think one major issue is getting people from the city to the farms during harvest time. It's a problem of coordination and transport.

Oh I know. I have a food forest, and aquaponics system, and some fence gardens. But I can't get rid of the grass (yet) because of the HOA (I did plant clover in it).

I'm just saying this would be a good space to start, if the robotics dev economics made sense: the market is there.

Nice how it is always other people's things that are silly. Lawns are silly in places like Las Vegas, but maybe not particularly silly in a place like Michigan or New Jersey.

My parents' NJ lawn needs a lot of maintenance -- they use a lawn service that mows the law every 7 - 10 days, fertilizes a few times a year, maintains the sprinklers, does some kind of regular weed control, and took care of a mole problem last summer.

Still seems silly.

Laws might seem to need a lot of maintenance, but unless you live in a desert and have the native vegetation that hardly grows(cactus, yucca, and mesquite like bushes), any landscaping besides a field of rocks will likely require more maintenance.

You could let your yard go back to the wild and that would be cool by me, but where I live people who replace their lawn go with mix of bushes, flowers, and trees with paths and other landscape features that require a lot more man hours in pruning, trimming, leafblowing, etc to keep looking good.

I don't live in the dessert, during the drought a couple years ago, our HOA replaced the lawn areas with some sort of low maintenance native bushes and flowers along with some sort of native grass that doesn't need to be cut.

So now instead of a weekly visit by a landscape crew of 3 to cut and maintain the lawn, we have a bi-weekly visit by a single person to do some light weeding and raking.

If you do it right, it can work out well. I just don't seem to see that in the East Bay where I live. I do my own gardening and cutting grass is easy. Pruning bushes and shrubs is a lot more time consuming per square foot in my experience but I would only mow once a month.

Yeah, I mean, in CT ours grew whether we wanted it or not. the choice was grass or trees, and its not great when trees grow literally right up to your house. if you wanted a buffer it either had to paved, gravel, or grass.

And paving in places with a lot of precipitation has some unpleasant side-effects.

You might be interested in working with groups such as Open Source Ecology https://www.opensourceecology.org/

I poked around a bit but wasn't able to find out what the nominal/design specs of the v2 robot are. Is this designed for slow movement through a farm field? What kind of payload, battery life, range, etc.? You've got all these videos and that forum, but a blog post about the project, your intentions and so forth, would be interesting.

The suspension and driving action are very impressive!

Thanks! Yeah there’s lots of missing pieces here as this is all a side project for me on top of a day job and social life (hah). I talk about my intentions in the “Rover Reveal” video (https://youtu.be/LIkX7wniYh0). This is for me an exploration in long service life 3D printed robotics as well as a software development platform for outdoor robotics.

You can also read an essay that explains my motivations for working on this kind of thing more broadly:

http://tlalexander.com/machine/

What material(s) are you printing the gearboxes out of currently? Nylon would give you a lot better longevity than most other thermoplastics, especially if you can SLS instead of FDM. It also looks like you're using straight cut gears in the current CAD files - you might want to try switching to a helical gear system if possible to increase torque and decrease wear.

If you can SLS, why not go with metal? Sintered metal gears (not 3D printed but "old-school" sintered) are basically the standard way to make gears for power tools etc. nowadays.

The cost differential is at least a full order of magnitude. Nylon SLS machines are available at a $10K price point. Metal (usually titanium) SLS machines are typically well over $100K. Cost of material is also usually an order of magnitude difference. The former is something you might see in a makerspace or college lab, but the latter is going to be restricted to corporate production environments for some time.

I assumed we were talking about someone buying gears from a 3D printing service. Then the cost increase for metal SLS seems negligible compared to the increase in robustness. If you're considering forking out $10K for a printer, OTOH, it means that you'll be running it for larger production runs, and then you might as well just get some molds made for traditional powder metal sintering. Especially when you factor in labour costs.

Very cool! Thanks for posting this. I've been looking for a fun 3d print / electronics projects. Thanks for posting videos too as it's much easier to quickly see what you're doing.

Thanks!

Reply on news.ycombinator.com