Most robot vacuums are a sealed box you rent from a company that keeps a map of your floor plan on its servers. A new project called oomwoo, from the maker behind Maker’s Pet, takes the opposite approach: open hardware, open firmware, open software, and a chassis you run off a desktop FDM printer. It maps the home with an inexpensive 2D LiDAR sensor, navigates itself using ROS 2 and the Nav2 stack on a Raspberry Pi, and integrates natively with Home Assistant — with everyday cleaning running entirely offline, no cloud account required. The hardware, firmware, and software are released under the Apache 2.0 license, and the developer is building it in public from the first commit.

That last part matters, so we’ll say it plainly up front: oomwoo is very early. As of late June 2026 it’s at the request-for-comments stage with no build instructions yet. This post is a tour of a promising project and what 3D printing makes possible — not a step-by-step you can follow tonight. We’ll flag exactly what exists and what’s still on the roadmap.

The division of labor

What you print vs. what you buy

The appeal of a project like this is the same one we point out in our 3D-printed drone guide: almost everything that isn’t a motor, sensor, or board can come off your print bed. The maker has confirmed the plan is to source motors, sensors, and the vacuum assembly from common suppliers (including AliExpress), and 3D print the enclosure, bumper, brackets, and dock. Here’s the rough split based on the project’s published design notes.

A convenience kit (motors, PCB, brushes, gaskets, LiDAR) is planned through Maker’s Pet, but the developer has been explicit that buying it is a convenience and never a requirement — every part can be sourced independently.

Under the shell

The reference architecture

oomwoo’s navigation runs on the same robotics stack used in research and industry, scaled down to a Pi. The compute choice isn’t locked — the developer is weighing a Raspberry Pi 5, an ESP32 running micro-ROS, or both — but the v0 milestone already covers a 3D-printed chassis, a ROS 2 Gazebo simulation, and LiDAR mapping with manual SLAM.

If you’ve read our rundown on how the maker community builds its own infrastructure, this will feel familiar: the same crowd that runs failure detection on a $30 Pi is the one that will happily assemble a navigation robot from open parts.

The number everyone asks about

What might it cost?

There’s no final bill of materials yet — the developer is targeting a rough BoM around mid-July and a finalized one toward the end of August. The figures below are the maker’s own stated targets, not verified build costs, and they’ll move as parts get sourced. Treat them as a direction, not a quote.

For perspective, that filament line is the same maker math we ran for the enclosure in our DIY Bluetooth speaker build: the printed shell is the cheapest part of the whole project, and it’s the part you can iterate on endlessly.

Why bother

The case for a vacuum that stays home

The strongest argument for a local-first design isn’t cost — it’s the cloud failures that have piled up around commercial robot vacuums. Reporting from Tom’s Hardware notes a string of them: at DEF CON 32 in August 2024, researchers demonstrated that several Ecovacs models could be reached over Bluetooth to access cameras and microphones, with one researcher describing the security as extremely poor. Hijacked DEEBOT units were later reported shouting slurs and chasing pets in U.S. homes, and a token flaw in DJI’s Romo line reportedly exposed roughly 6,700 vacuums, their floor plans, and live feeds. One owner went as far as reviving a remotely bricked vacuum with custom boards and Python to run it offline.

oomwoo’s reference design sidesteps that whole attack surface by navigating on 2D LiDAR and bumper sensors — nothing pointed at the room. The usual alternative, the well-regarded Valetudo project, takes local control back from a vacuum you already own, but Tom’s Hardware notes that on many models it requires rooting the firmware, which can mean disassembly, a voided warranty, and a change that can’t be undone. Building from open parts skips the rooting entirely.

Where we come in

Printing the chassis in San Diego

We’ll be honest about our role here: Dreaming3D isn’t affiliated with oomwoo, we don’t sell the kit, and we haven’t built one — the files don’t exist yet. What we can do is the part we do every day: print a clean, durable chassis once the design files land, in a material that suits a robot that bumps furniture and lives on a warm floor.

Material picks for a floor-roaming robot

A vacuum chassis takes knocks, sees occasional warmth, and benefits from a little stiffness for the LiDAR mount and motor brackets. PLA will technically work for a first prototype, but for a part you actually run, PETG is the sensible default — tougher and more heat-tolerant — with ASA worth considering if the unit ever sits in sun near a patio door, and a carbon-fiber composite for bracket stiffness. Our 2026 filament guide walks through the trade-offs in depth.

And if the project takes the retrofit route the developer floated — mounting a Pi, LiDAR, and a replacement I/O board into a vacuum you already own — that’s squarely the kind of work in our scan-to-print and reverse-engineering wheelhouse: measure the original part, model a bracket or placeholder, print it to fit.

Questions makers are asking