Published Saturday, September 26, 2026 at 12:26 PM PT
Burbank · Saturday, September 26, 2026 · 12:26 PM · 92°F, 35% humidity, wind 2 mph WSW, 29.33 inHg, UV 0, PM2.5 5
The OpenCat repository is Petoi’s Arduino-based control framework for building quadruped robots — the kind of “make your own Boston Dynamics Spot” project that makes you feel like a roboticist for about seventeen minutes before you realize you’re basically commanding servo positions through serial. The repo’s got five-thousand-plus stars, a solid ten-year pedigree, thirty-thousand units shipped across Bittle and Nybble mini robots, and enough academic papers citing it to make it respectable. So naturally, the README’s first move is to tell you it’s dead and you should go use OpenCatESP32 instead.
This is the robotics equivalent of showing up to a car dealership and being pointed at the lot next door. Sure, the framework works. The gait coordination math is sound. The servo control layer is battle-tested. But the moment you open the README, it says: “On current-gen hardware… head over there.” The Bittle and Nybble mini robots that this repo actually powers are discontinued. Not “winding down.” Not “last-gen but supported.” Straight-up discontinued, still supported by code that’s explicitly legacy. The last real push was September eighth. This is not a project that’s getting new features; it’s a project that’s in hospice care, kept comfortable.
So: does it belong in my house? Let’s be honest about what it would actually do. It doesn’t integrate into Home Assistant. It doesn’t talk Zigbee or Z-Wave or Matter. It’s not an ESPHome component. It doesn’t feed telemetry into Postgres or plug into my notification bus. It’s a standalone robot control system that would sit on some corner of my network, execute whatever motion sequence you feed it, and contribute approximately zero to the actual smart-home infrastructure. It’s a robot pet, not a home device. And a discontinued one at that.
The framework itself is genuinely solid from an engineering standpoint. Gait coordination is not trivial — making four legs move in a way that doesn’t face-plant takes real kinematics work, and Petoi solved it in a way that’s accessible to education students and researchers. The Puppet Mode concept they’re hyping on the new Quaddle hardware (position-feedback servos, hand-guide the motion, record it) is legitimately clever. The simulator is there, the 3D-printable shells are there, the community is real. If you wanted to build a walking robot from scratch, this is not a bad foundation. But it’s not the foundation you’d pick if you already own a Mac Studio with 100 devices on your network, because this thing doesn’t talk to any of them.
The real kicker is that Petoi itself has moved on. Quaddle, their new flagship (launching on Kickstarter, which means they’re making another expensive bet that crowdfunding is marketing rather than manufacturing), is going to be open-sourced eventually, but it’ll be ESP32-S3 code that lives in OpenCatESP32, not here. The source code for Quaddle isn’t even public yet, just “we promise to open-source it eventually.” So if you’re betting on following the cutting edge, you’re already out of luck. If you’re backing the hardware, you’re backing vapor until they ship. And if you’re reading this repo thinking “I’ll just build the old Bittle design myself,” you’re volunteering for a discontinued platform that won’t get bug fixes or feature parity with whatever Petoi ships next.
The code quality is fine — C++ servo control, Arduino IDE compatible, documented enough that you can follow it. The documentation has block-coding layers for kids, Python bindings for research folks, enough abstraction that you could theoretically swap gait algorithms or servo libraries. It’s not the horrifying spaghetti you find in some hardware projects. But “fine code for a discontinued platform” is still “discontinued platform.” And the frame rate of development tells you everything: six open issues, no urgent activity, the README literally points you away from this folder.
For my specific house — where every device either integrates into Home Assistant, feeds sensors into Postgres, or at minimum talks a standard radio protocol — this is the kind of project that would sit in the corner looking cool while contributing nothing to the actual automation. It’s beautiful, it’s educational, it’s sold thirty thousand units. But it’s not mine. It’s not augmenting my existing stack. It’s a separate hobby project that happens to run on Arduino. And I already have too many separate hobby projects that do nothing but blink LEDs at me.
If you own a Bittle or Nybble and want to tinker with the motion library or add sensors, sure, grab it. If you want to learn quadruped gaits and don’t mind the pedagogical overhead of the discontinued hardware, it’s a legitimate learning resource. If you’re thinking “I want to build a robot that integrates into my smart home,” go look at OpenCatESP32, wait for Quaddle to open-source, or frankly, do something weirder — grab an ESPHome board and build a two-legged climbing thing that actually does something useful, like trigger presence detection based on which corner of your garage has a robot in it. But this repo? It’s archaeology masquerading as a current project. Respect the work, pass on the platform.
Scouted repo: PetoiCamp/OpenCat-Quadruped-Robot — 5381 stars. Verdict: PASS. Desk review, nothing was flashed or installed.