back
80 comments
GitHub page is better & cleaner for those curious: https://github.com/PetoiCamp/OpenCat
Yes. The Petoi site boasts about how many stars they have on Github, but don't actually link to it.
Petoi had a rep at OpenSauce in San Francisco. I picked up a Nybble, because I like the look of Nybble better than Bittle.

It's a fun little toy and I really appreciate what Petoi are doing with the robots & the OpenCat framework! However... due to either a design flaw or my own subtle mistake in assembling it (haven't identified which yet), the poor cat's tail is too long! It hits the floor and topples the bot over!

Mod yours into a bobcat or a lynx!
Charles Stross’s Accelerando comes to mind.
Very much so! Someone clearly did not (or did?) read the book...
I have Bittle, their Open Source robot dog. Got it when they were doing the kickstarter.

It’s pretty cool :)

Despite their cool look and visually appealing design, QUGVs are still not optimal for practical real-world applications. Their limited mobility in comparison to UAVs for ISR operations, coupled with a mediocre payload capacity relative to traditional UGVs, results in a platform that have disadvantages from both categories. In my opinion, QUGVs are primarily beneficial for educational purposes or just a show or a ‘wow’ reaction from those outside the robotics field. For example, Boston Dynamics cannot descend stairs in reverse last time I checked, while Unitree tend to overheat rapidly, with internal components that are susceptible to failure, and they are constructed with cheap components, such as three Jetson Nanos, rather than a single, more robust SBC, etc.
You're not wrong, but probably applying a critique which is irrelevant to the product under discussion. I don't think potential owners of 'robotic pets' will be making purchases based on maximising payload capacity, or practical real-world applications.
Most analog cats won't descend stairs in reverse or carry loads either.
They can manoeuvre uneven terrain in areas where UAVs and wheeled UGVs are unsuitable though. Combined with UAV capabilities, it could be a very capable mapping or ISR tool in places like caves, or following earthquakes. The ‘hopping’ type in particular might be particularly valuable for lunar and asteroid mapping.
Humans bred cats for thousands of years for utility that had nothing to do with "payload capacity"
> Their limited mobility in comparison to UAVs for ISR operations

UAVs are highly susceptible to wind and drift, especially indoors with no GPS, and while DJIs collision avoidance is pretty good, any land based vehicle won't (fatally) crash if it touches anything.

If you're looking for a visually appealing design, then it's hard to beat a real cat. The internet agrees.
QUGV? UAV? ISR? SBC?

OMG, WTF!

With large language models, it should be possible to make robot "pets" that can talk. You won't fit a big graphics card into one of these things at the current SOTA, but you don't need to. It would just need to be internet connected and calling APIs.
Yeah - and now the APIs can do vision too.

About time to build Teddy from the movie "AI".

I am eager to see a wheeled car like that support an [old] mobile phone over and is cheaper than USD 30? Good for education. The "prototype" via Lego Mindstorms is more expensive.

I think much of the cost is a good motor now? There were startups trying to make this but I think they are closed [1].

[1] https://web.archive.org/web/20190104173241/http://www.simple...

FYI, Romotive became Zipline, and is now (one of the) most successful drone delivery companies. It's a wild pivot story.
It's amazing what's available commercially right now. So quick after Boston Dynamics shown and sold theirs for the first time.

https://unitreerobotics.net/robotdog/unitree-go1-pro/

You could get a decent computer and a robot dog for a price of Mac Pro.

Super neat. I am slowly trying to get my kid into the space so I will be watching this one.

Incidentally, if anyone can share similiar efforts or kid ready sets, I would love to hear about it.

Depends on their age and what level, there are “toys” you can buy with some visual programming like DJI robomaster, or advanced ones that actually use real lidar and stereo cam with ROS like AgileX limo (original price is $1000 btw, if a reseller is selling it for more they are just gauging the price), and also there are DIY ones that require 3D printing and such.
What are the pros of this compared to the seemingly cheaper Sunfounder PiDog? The latter seems to be a bit better wrt sensors and the like.
Cute until someone straps a gun to it
So, just like a normal pet?
What I need is a solar powered turtle that can silently mow my lawn.
Ä-Ä-Äye kän h-h-h-äz chee-chee-zz-zz-byrrgr-r-r-r-r?
>Customizable parts. Open choice of components.

He says, gesturing at the robot's groin.

Both the website and the project's Github repository loudly trumpet that they are "open source" next to photos of proprietary hardware.

In https://github.com/PetoiCamp/OpenCat-Old/issues/7, the creator admits that the "nyboard" (https://docs.petoi.com/nyboard/overview) at the heart of their robots is derived from Arduino, but insists that they don't need to comply with the licensing terms for derivative boards (https://support.arduino.cc/hc/en-us/articles/4415094490770-L...) because Arduino is unlikely to sue them.

It's unfortunate that the hardware isn't open source, but I think he's technically right that his board is not an Arduino derivative from a legal perspective.

"I initially built the prototype with an Arduino board. Later I got my friends designing the NyBoard, a customized PCB based on ATmega328P. We will release the BiBoard, another board based on ESP32. They both support the Arduino IDE but are not using Arduino's brand"

Not “open” like “open source”. “Open” like “OpenAI”

“As we get closer to building AI, it will make sense to start being less open. The Open in openAI means that everyone should benefit from the fruits of AI after it’s built, but it's totally OK to not share the science. — Ilya Sutskever https://openai.com/blog/openai-elon-musk

Presented with a signup popup modal before I could even scroll down to read the first paragraph. I get you want engagement, but that's too aggressive for me. Page closed.
Do you feel like this focus on minutia versus substance has led you to miss otherwise insightful content?
Real pets looking nervously around…

In seriousness this is really cool! I wish I had this in school for robotics learning.

The price to adopt a real cat is ~100usd in most states. Why not just get a cat? This projects like 300usd.
Total cost of ownership of a cat is considerably more than $300, unless you plan to make your cat hunt its own food, fail to license it, and never take it to the vet.
Because when you take a real cat apart to see how it works you can never get everything back inside and it never works quite right again.
Real cats eat and shit and need medical help at times, while this doesn’t.
Superior interfaces.