I have to ask. Has anyone on the Yarbo development team ever actually used the snow blower feature, made snowbanks, or even seen snow? Some of the solutions are so baffling that, especially when it comes to the software, a question like this inevitably comes to mind.
I know there are testers in snowy regions, but are there any actual developers?
Having led software and product development projects myself, I’d argue that if the software is being developed by people who don’t actually deal with snow themselves, it’s about as effective and successful as teaching speech to someone who is congenitally deaf. It’s possible, but extremely time-consuming and highly inefficient.
Typically you see pictures or short clips of them in what looks like an underground parking garage with snow from a snow machine. With some of the software decisions I’d also lean towards many of the programmers and/or development leads not actually understanding real snow conditions. It is workable as is but very frustrating at times. Lots of good tips and tricks here and in the Yarbo wiki though.
That’s what Bryan is for. lol. He keeps taking the points for most posts. Make him work for them. 
Jack
Please rest assured that we do conduct real-world testing. At the same time, there may still be scenarios we do not encounter during testing, and there are also some issues we are continuing to optimize. We truly appreciate your ongoing feedback, as it helps us continue improving the product.
I said the same thing maybe a month ago. I don’t think the developers ever used a snowblower before. I’d love to hook up sensors to me and my walk behind to help train their model. It’s utterly horrible.
Yes, I know that Yarbo does real-world testing. But that is exactly what I’m referring to. If the programmers themselves do not know what snow is like and how it behaves, you first need to teach them, and then create about a thousand tickets before they even begin to grasp what is going on. Instead of that, you could use developers who have grown up and lived with snow and dealt with it their whole lives.
It is a completely different thing. As I said, trying to solve it this way is about as efficient as trying to teach a deaf person to speak using only notes and tickets.
If you truly want a product that works and delivers good results, hire developers directly from regions where there is real winter and proper snow. Or at the very least, organize a workshop month where the developers go out themselves and fight with their devices in the snow.
If your operating model is that you have testers in snowy regions while developers who don’t understand the conditions receive tickets from them and try to understand the problems based on those reports, you will never end up with a product that actually works in the real world. Or at least not in time — the product will become obsolete before that happens because someone else will do it properly.
The same applies to mechanical designers as well. Which product development version is the Yarbo Core 2025 already? At least the third or fourth if you count the Snowbot days.
And it still has completely fundamental basic design mistakes that would be obvious to any engineer who has actually done snow work — things that you simply cannot design that way.
Starting with things like the extremely low ground clearance, sharp edges left underneath the machine, and the rails on the underside that the zero-turn logic is desperately trying to fight against.
I find it hard to believe that the person who designed that structure has ever actually dealt with real snow work or snowblowers. And if they have, then they clearly haven’t really thought about why things in those machines are designed the way they are.
We’re very sorry to hear about your disappointment. Yarbo is still in a stage of continuous development and improvement. However, we want to assure you that every new firmware release goes through hundreds or even thousands of real-world tests, as well as internal PPP beta testing, ensuring the scale and rigor of our testing process.
In terms of team background, a significant portion of our product design team has studied in North America or lived in snowy regions, so they are quite familiar with snow conditions. In addition, we have built a dedicated snow testing facility to evaluate the machines. We also conduct annual winter testing programs, and regularly organize trips to heavy-snow regions for field testing and training. In fact, our R&D team will be traveling to Mohe—a heavy-snow region—tomorrow to conduct further testing.
Some of the issues you mentioned have already been reported to our R&D team and are currently under evaluation. As soon as we make progress, we will share updates with everyone right away.
Sounds bitterly cold
, seems like a good testing area for Sodium Ion and/or hybrid sodium/lithium battery testing
I think one reason why some people have excellent results with the snowblower while some people think it’s awful is due to the large amount of variability in snow and snowy conditions.
It’s probably not that the dev team isn’t familiar with snow. It’s more likely that they just weren’t initially aware of every possible scenario that the Yarbo bot might encounter. Additionally they likely targeted their initial releases at the 80-90 percentile of scenarios and now need to develop strategies for the next largest percentile group.
Example: My Yarbo snowblower has worked extremely well for me this winter. Once I learned how to optimize its zones and parameters, it became extremely reliable. But @rcguymike - who lives in the same state as me - still has a lot of reliability issues. Why?? His part of the state gets a completely different quantity and type of snow compared to what I get. What works for me won’t work for him because the use case is different.
So the dev team just needs to continue to capture edge cases and develop strategies that accommodate them. Yarbo won’t ever be perfect of course, but I think it’s reasonable to assume the iteration process will continue and that they will keep chipping away at the edge cases that most users are experiencing.
Ask the Inuit how many kinds of snow there are. Maybe they should visit them for a season? LOL! In all fairness, I think it is hard for a controlled environment to emulate every type of environmental variable and type of precipitation, freezing, thawing, re-freezing, etc.
For example, take a look at the big snow storm we had just recently. Fluffy dry snow topped by an ice storm. My Yarbo with the controller and lawn tracks did fairly well until I got bold and high centered it by trying to do things with tight turns or drive up a driveway entrance ramp that was not flush with the roadway. Straight snow blowing was awesome until the ice storm started in the afternoon. The next day, we had an inch or so of very hard ice. I couldn’t stomp through it with my feet. Buzz LightYar looked at me like, you’re kidding, right? Even with my TroyBilt SnowZilla, I couldn’t attempt anything with that stuff. I had to chop it out with a floor tile scraper tool a few inches at a time. It’s a robotic snow blower, not a farm tractor.
Itching to get out and mow with my Yarbo,
Kevin & Anita Schlosser
Parents to Yarbo: Buzz LightYar
Expecting another soon.
I think your question is terribly unfair to the development team, hardware and software. If you have never designed/redesigned a blended software and hardware project, you don’t grasp the limitations you have to design around. Only some of them are edge cases and then there is scope creep. I am leading a product launch myself.
There is also intelligent use of the Yarbo itself. I use it to clear a 250’ driveway with no slope. The Yarbo handled that with my 18” of dry snowfall and 6” blowing snow followed by slush. It isn’t magic, the 24” berm of salty, icy chunks the county snowplow left behind were pure torture. Spade > snow shovel > Yarbo, was the right order for choices there.
Were I to make a criticism, it is the selection of pathing algorithms. Certainly, the Yarbo is not offering the path I was taught for effective snow blowing, the “center‑out racetrack” which works well on level areas. Steep driveways are their own special evil in snow. Been there, done that, with a snowblower. You have to tile out the driveway and think carefully about pathing. Visualizing a Voronoi decomposition is helpful, but not for everyone.
I believe the software and hardware (and support) teams are operating with best intentions. I do understand where you are coming from as I’m frustrated too. But, I see a lot of potential.
We’re getting hammered (with a blizzard) tonight, and Yarbo couldn’t do it. It kept zero turning, getting stuck, escape routine, end up in No-Go, end up stuck, end up high-centered (and not notifying me? this seems new it just spins the auger and runs the battery). So frustrating…
!!However!! - when I took over with the controller from the comfort of my warm house I was the only one in the neighborhood clearing their driveway in the blizzard. Yarbo was not getting stuck and was a champion. Still way better than being outside and pushing the snowblower up the hill!
I suspect that’s where the vision for M series makes sense - there is both demand and the work areas are less varied (typical suburban lot). Good luck to them. M series won’t be able to handle my driveway.
The Yarbo hardware and software teams try to balance product goals and constraints, tech feasibility, customer feedback, and resource constraints in a continuous struggle for which will be dominant in the next product cycle. I have experience on both sides (software + hardware) as well. I have 250’ (76.2m) and serious slope (most plow operators refuse to plow).
More often, company beauracracy (i.e. org structure) tempers excitement - so even if the devs are very passionate and capable, they may not get to solve things in the most obvious or transparent ways to customers. Understand I am not providing an excuse - I am stating the dev teams aren’t unilaterally defining and designing. EDIT: to your point, with the right expertise, there may be better results if they were!
However, @Yarbo-Forum what you are seeing is customer frustration because their JTBD is not being met. This could be solved multiple ways, but I propose my view of the highest value below.
Enable a robust and deep API!!
It’s coming I hear, have a spec yet for preview? I am sure you have a line of us eager to beta-test. Count me in!
I believe the hardware itself is very capable. However, the variable conditions (ground, weather, rando parked car, ad hoc) that the Yarbo software team is trying to program for are too numerous and will only result in frustrated users. Moreover, users waiting for a release cycle of perhaps 1 or 2 releases per season to maybe get the fixes applicable to them - yields mixed results. It is not reasonable to believe yarbo will meet everyone’s needs in a timely manner given the aforementioned constraints. Yet, customers have real needs that can’t wait until Yarbo is robustly tuned.
If it were me focusing on the Y series, I would have the software team focus on opening up the module controller API, the core API, then give customers fine grain detail over their workplans much like 3D printer (e.g. Orca Slicer) software does, then create composable routines we can select from (180 zero turn w/head fully raised, 180 soft arc with head operating, recharge, etc). When I build the workplan, I select the default pattern overlay, 180 soft arc here, start in middle, go clockwise over there, raise the head for this spot due to a DIY asphalt patch, if stuck here - use stuck routine 3, if stuck there - use stuck routine 6, etc…
@Yarbo-Forum For hands-off automation I recommend focusing on your hardware, open the APIs, and develop the tools to enable and compose our own routines.
This path will lead to more customer promoters and their really jealous neighbors.
This is the statement I want to tell my neighbors:
“Yesterday morning while drinking coffee I vibe coded a new yarbo workplan because we have lake effect snow coming and my uncle always parks his car in a way that blocks the front steps. I told Yarbo to throw all the snow onto his car so he gets the message…”
Thank you for sharing such thoughtful and sincere feedback, and also for your understanding of the many factors we need to balance.
We understand that API integration is something many users are looking forward to, and it is a direction we are actively considering. At the same time, there are many factors involved in making it available in a reliable and useful way, so it may still take some time.
We truly appreciate your patience, your detailed suggestions, and your continued support. Feedback like yours is very valuable to us as we continue improving the product.
Well, since I regret my purchase decision more & more with each snowstorm, I’m going to go out on a limb and say no, they have no idea what snow is.. But hey, I’m probably wrong..
I mean, what “automated” piece of equipment commits a slow suicide after tapping a car tire. I’ve hit something now I must die…
Hi. Can you share what happened with your Yarbo?
You said it bumped into a car tire and then what happened?
Maybe we can help.
Well, anytime it brushes pretty much anything, it goes into “collision” mode and has to be reversed a foot and resumed manually. If not it will sit there and drain it’s battery to 0 in the 15 degree weather. Question, why can’t it just back itself up a foot and resume? I usually start it before bedtime and if I don’t hear the collision warning while I’m sleeping, I wake up to a deceased Yarblo in the driveway. I then have to try and drag it back to within extension cord range, and resuscitate it, in the 15 degree weather. It really should be able to complete a plan without human intervention. I don’t think mine has ever completed one without needing something.
Good morning.
Yeah, I understand exactly what you’re describing.
My first thought is - It’s very important to prevent Yarbo from bumping into anything. Mostly because it is a very large powerful and heavy robot. It can easily cause damage if it bumps into something.
So if it’s currently possible for your Yarbo to bump into a vehicle, the physical location where that vehicle is parked should be either marked as a no go zone, or removed from the driveway’s snowblowing work area.
At my house, my two kids have a designated spot where they park and charge their cars. That location is NOT a part of the normal snowblowing zone. But I created a secondary snowblowing zone just for that location, and I select it only when the kids’ cars are not parked there.
Hope this makes sense and is helpful. Let me know what you think.
This^, plus do you have car detection enabled? It is still beta but does work some of the time. Best to use no-go zones and turn them on and off as needed.
You should visit the Keweenaw Research Center in Houghton County Michigan like the car companies do for cold-weather testing.