THE PIXAR LAMP IS REAL - 260 DOLLARS IN PARTS AND IT MOVES AND TALKS BACK
A desk lamp that nods, stretches when it wakes and chats with you - open-source hardware and code inspired by the Pixar lamp and Apple's ELEGNT research.
by Binh Pham / Human Computer Lab
RoboticsHome
Built withRaspberry Pi3D printing
- difficulty
- ●●●○○
- time
- a long weekend
- license
- GPL-3.0
- repo
- repo ACTIVE587 stars
●●●○○ · a long weekend · GPL-3.0 · 587 stars · repo ACTIVE
WHAT YOU’LL NEED
Jump to section
COMPAREE VERDICT
LeLamp is an open-source robot lamp from Human Computer Lab, inspired by Apple's ELEGNT research on expressive movement and the classic Pixar lamp. Five STS3215 serial-bus servos give it a five-axis arm, a Raspberry Pi with a ReSpeaker HAT handles the microphone and speaker, a 24-LED WS2812B matrix in the head shows expressions, and a camera is part of the hardware. The runtime lets you record and replay movements (it ships with moods like curious, excited, nod and wake-up) and includes a conversational voice agent built on LiveKit and OpenAI, so you need your own API keys and an internet connection for that part. The authors put it at about 260-310 dollars and 8-12 hours, rated intermediate to advanced: you will solder, set servo IDs, wire through a stripped Ethernet cable and set up Linux over SSH. Stick to the STS3215 servos - other servos mean rewriting the control code. A charming, well-documented build, still marked early development.
IN THE REPO
GOOD TO KNOW
- —3D-printable parts as .3mf files (8 prints) plus the full OnShape CAD
- —Python control code (motors, record/replay, LEDs, voice agent) in the separate lelamp_runtime repository
- —Bill of materials as a Google Sheet with US reference links (mostly Amazon) and a total of about 311 dollars; the team says the links are reference only, so buy locally
- —Assembly guide with photos and wiring diagrams in the README
- —GPL-3.0 licence — allows commercial use with source disclosure
- —No PCB design included; servos and electronics are off-the-shelf modules wired together
Parts to buy
8 itemsFrom our check of the build. Exact quantities and part numbers are in the creator’s BOM.
Can I build this?
Build at your own risk. Projects involve tools, electronics and sometimes mains voltage — follow the creator’s safety notes.
Videos
Building A Real Life Pixar Lamp
The original designer's build video; more clips on the official LeLamp YouTube channel.
More builds like this
All projectsGallery
Start here
Navigation into the creator’s own docs — we don’t rewrite the guide, we route you to the source.
- 1.Read the full README on GitHub (The README contains the complete bill of materials, wiring diagrams and software setup instructions. Start there to confirm you can source the servos.)
- 2.Print the 8 parts from the 7 .3mf files in /3D/(PLA works (PLA+ recommended), 0.2 mm layers, tree supports, and a bed of at least 200 x 200 x 150 mm.)
- 3.Order the servos and electronics (The BOM links are for reference only - the team says to find the best deals in your region. Make sure you get genuine STS3215 serial-bus servos, since other servos need different control code.)
- 4.Assemble the mechanical structure (Follow the photo guide. The servo horn alignment matters — mark the zero position before tightening screws.)
- 5.Install the Python environment and dependencies (Clone lelamp_runtime and install it with uv (`uv sync`, plus `--extra hardware` on the Pi). It pulls in LeRobot, the Feetech servo SDK and LiveKit agents; put your OpenAI and LiveKit keys in the .env file.)
- 6.Test each subsystem individually(Use the runtime's test scripts to check the five servos, the audio and the LEDs, and libcamera-hello for the camera. Fix one thing at a time before running the voice agent.)
Resources
Documentation, files and community threads for this build — we link straight to the original sources and never rehost the creator’s files.
KNOWN ISSUES
- The servos are the single biggest sourcing decision: the BOM's links are US reference links only, so find genuine STS3215 serial-bus servos from a local supplier. If you buy different servos, you will need to rewrite the serial protocol code unless they use the exact same Feetech STS command set.
- The voice agent needs a stable Wi-Fi connection and your own OpenAI and LiveKit accounts - without them the lamp can still move, record and replay, but it will not talk.
- Each servo needs a unique ID (1-5) set with the motor setup script before assembly - label them first, because mixing them up after they are built into the arm is painful.
- Calibration matters: put the lamp in its zero position and follow the guide exactly - do not fully rotate the two yaw servos (base and head), only about 90 degrees each way, or you can damage them. The built-in moods (nod, idle, wake-up and more) are recorded movements you can re-record to suit your build.
- Voice interaction depends on the cloud agent and your microphone setup; the runtime applies LiveKit noise cancellation, but a loud room can still confuse it - the agent is prompted to ask you to repeat yourself when audio is noisy.
Can I use different servos if I cannot source the STS3215?
Yes, but only if the replacement servos support the same Feetech STS serial protocol. If you switch to a different protocol (e.g. Dynamixel, plain PWM), you will have to rewrite the servo control layer in the Python code. The repository does not include alternative drivers.
Does the camera or microphone capture anything permanently?
Movement recordings are saved locally, and the voice mode streams your speech to LiveKit and OpenAI for the conversational agent. If you want nothing to leave the device, don't enable the voice agent.
Can I run this on a Raspberry Pi 3 or a Pi Zero?
Yes. The project's bill of materials lists a Raspberry Pi 4 and says a Pi 3B, 4 or 5 can be used, and the lamp base is designed to hold a full-size Pi with the servo driver. The heavy lifting of the voice agent runs in the cloud (LiveKit and OpenAI), so you do not need a powerful board.
What is the power consumption?
The docs use two 5 V 2 A supplies: one with a DC plug for the servos and a USB-C charger for the Raspberry Pi. You can run everything from one supply by feeding the Pi from the motor driver, but the team warns this risks bricking the Pi.
Can the lamp talk back?
Yes - the head has a speaker, and the LeLamp runtime includes a conversational voice agent built on LiveKit and OpenAI's realtime model. You supply your own OpenAI and LiveKit keys, and the conversation runs through those cloud services.
Community builds
No community builds yet — be the first, we feature the best ones.
Discussion1
FROM THE COMPAREE TEAM
Five servos, a Pi Zero 2 W and some printed parts, and you have a lamp that moves and talks. If you have built one, which part took longest: printing, servo setup or the voice runtime?
Binh Pham / Human Computer Lab
Binh Pham designed LeLamp's mechanics, electronics and software runtime while at Human Computer Lab, inspired by Apple's ELEGNT research on expressive robot movement and the classic Pixar lamp.
DISCLAIMER
- Comparee is not the author of the projects featured here. All rights to each project belong to its creator — every page links to the original source, and we never host creators’ files.
- Information is provided without warranty and may become outdated as projects evolve. Prices are indicative bands only — always check the creator’s parts list for current costs.
- Building and operating any project is at your own responsibility. Protective equipment, safe workshop practice and compliance with local regulations are the builder’s responsibility.
CompareeTEAM2mo agoedited
Practical notes from our verification: the repository was last updated in August 2026, the README links a full build guide (prerequisites, printing, servo setup, assembly, runtime setup, control and common issues), and the bill of materials is a Google Sheet. The authors note that its purchase links are reference only and that you should find the best deals in your region. The lamp uses five STS3215 servos, a Raspberry Pi Zero 2 W, a camera, microphone and speaker, and 24 LEDs, all on a 5 V supply. The conversational voice agent runs through the separate lelamp_runtime repository and needs LiveKit and OpenAI keys, so factor in those accounts. The guide rates the build as intermediate to advanced at about 8-12 hours, and if you get stuck there is a troubleshooting page, a Discord and GitHub issues. Note that the team says it is also working on a closed-source version. Correction (4 October 2026): we re-checked this page line by line against the project's own repository, documentation and videos, and fixed errors in earlier versions.