StackChan AI Desktop Robot (ESP32-S3)
AI desktop robot using ESP32-S3 CoreS3 with 2.0-inch touch screen, 0.3MP camera, Wi‑Fi/BLE, dual mics, speaker, servos, NFC, LEDs, and 550mAh battery.
127+ curated AI devices, each tagged with its original open-source repository and official buy link. Spanning M5Stack, LilyGO, Seeed, ESP32, RISC-V, Raspberry Pi and the global open-hardware ecosystem.
AI desktop robot using ESP32-S3 CoreS3 with 2.0-inch touch screen, 0.3MP camera, Wi‑Fi/BLE, dual mics, speaker, servos, NFC, LEDs, and 550mAh battery.
The category icon — a multi-tool with thriving community firmware.
The big 10.3" TRMNL: sharper e-ink, a gesture bar and a magnetic charging dock.
Round touch knob with RFID — an ESPHome kitchen-timer favorite.
Upgraded Cardputer: IMU, better audio codec, 3.5mm jack and a 1750mAh battery.
StickC successor and Resident's "start-here" kit.
Camera, mic and touch screen in one core unit.
Productivity device with a 72×16 LED screen for custom busy statuses, Pomodoro timer, and smart home/API integrations.
Pocket stick with screen, IMU and IR — a security-firmware staple.
4.7" touch e-ink tablet.
5" 720p tablet on the new ESP32-P4.
Hackable 7.5-inch ePaper display with soft-touch ABS enclosure, steel kickstand, and rechargeable battery.
Tiny voice puck — the default Home Assistant voice sample.
Thumb-size enclosed puck with a 0.85" screen.
Watch-form AMOLED with mic and speaker — a DIY voice-widget platform.
The ESP32-2432S028: a 2.8" touch TFT with a 4,000+ star community — the on-ramp gadget.
4.2-inch e-paper note display with ESP32-S3, Wi-Fi/BLE, voice input, speaker, mic, NFC, buttons, USB-C, and 2000 mAh battery.
ESP32-C6 development board with 2.16-inch 480×480 capacitive AMOLED, Wi‑Fi 6, BLE 5, IMU, RTC, and dual microphones.
A fully assembled 4.2-inch four-color e-paper development device with ESP32-S3 and open reference firmware.
Round 1.75-inch 466x466 AMOLED touch dev board based on ESP32-S3, with optional GPS and case variants.
$27 open smartwatch running InfiniTime.
Consumer pixel clock that everyone reflashes with AWTRIX 3.
56-key QWERTY pocket computer — the vibe-coding darling with a 700+ star awesome-list of community firmware.
Round touch display with an IMU.
The V3 successor: 28 dBm high-power LoRa, solar charging and a GNSS port.
2" touch core unit — the classic M5 platform.
A digital art frame that displays AI-generated images and can transform your photos into artwork.
Enclosed voice-assistant kit with offline speech recognition.
Official Home Assistant voice puck, fully reflashable.
7.5" mono e-paper terminal with roughly three months of battery.
The T-Deck plus GPS, a 2000mAh battery and a case — a near-complete handheld out of the box.
The e-paper T-Deck: 3.1" touch display, QWERTY keyboard, GPS, LoRa and an optional 4G modem.
USB-A dongle with a 0.96" LCD.
Knob, screen and sub-GHz radio — the Bruce/Flipper-adjacent one.
Open e-paper watch — flash your own faces.
JavaScript smartwatch with an app loader.
1.54" e-ink mini unit.
7" HMI with a swappable radio module.
4" square in-wall panel — a Home Assistant favorite.
Open-source AI glasses with a color display, dual bone-conduction speakers, dual mics, and a low-power on-device AI processor.
7.3" six-color Spectra e-paper terminal.
The workhorse e-paper gadget for calendars and transit boards.
BlackBerry-style keyboard with a trackball; optional LoRa makes it a favorite Meshtastic handheld.
Handheld ESP32-P4 board with 4-inch TFT/AMOLED touchscreen, ESP32-C6 Wi‑Fi 6/BLE, camera, SX1262 LoRa, mic, speaker, and L76K GPS.
1.91" AMOLED — a vivid mini dashboard.
Low-power e-ink Meshtastic node, fully enclosed.
Thumb-size gadget with a 0.85" display.
Touch dev smartwatch with a LoRa option.
Rugged dev smartwatch with a 2" AMOLED, LoRa, GNSS and NFC — the T-Watch S3 successor.
The default Meshtastic node, with a 0.96" OLED.
Wearable Wi-Fi-testing gadget.
Capacitive-touch "CYD 2.0" — what people usually mean by CYD these days.
3.5" capacitive touch panel — "the better CYD".
4.3" capacitive touch panel for dashboards.
9.7" recycled e-reader display.
Unbranded 15-key LCD keypad module with 72×72 pixel keys and USB-C for embedding in custom hardware.
1.9" IPS ticker and gauge board, with shell and battery variants.
A miniature voice conversation kit with a color screen that integrates the AtomS3R main controller, a microphone, and a speaker into a single unit. It can connect to cloud voice assistants, making it ideal for building desktop chat terminals.
Palm-size tri-color e-paper.
Recycled Kindle screens, programmable with Arduino or MicroPython.
Color e-paper variant of the Inkplate 6.
10.3" touch e-paper terminal with 16-level grayscale and up to six months of battery.
13.3" six-color Spectra e-paper terminal — the big digital-frame one.
7" wall-dashboard panel.
GPS, LoRa and an 18650 holder — the go-to mesh node.
A palm-sized voice development device with a built-in microphone and speaker. It supports local wake-word recognition and can connect to cloud services for voice conversation.
QWERTY pager with a 2.33" touch screen, GNSS and LoRa.
ESP32-S3 dev board with 1.54-inch 240×240 LCD, dual mics, speaker, IMU, TF slot, Wi‑Fi and BLE 5, with optional touch and battery.
7.5-inch four-color e-paper display device with 1800mAh battery, upgradeable to 2500mAh.
Screen-and-buttons HID payload gadget.
2.9" e-ink Meshtastic terminal.
E-ink plus LoRa node.
3.5" HMI panel with enclosure options.
An edge AI computing host housed in a transparent pyramid enclosure, capable of running adapted large language models and vision models independently on-device. It features a built-in microphone array and speaker with Ethernet connectivity: shown here is the 4GB RAM version.
An edge AI compute kit equipped with Ethernet and Wi-Fi, capable of deploying adapted vision, voice, and small language models. Microphones and speakers must be connected externally, making it well-suited for building local inference prototypes.
A standalone Linux vision device with a built-in camera and microphone that runs image recognition algorithms offline locally. It transfers processing results via USB or Wi-Fi, ideal for quickly prototyping small vision projects.
A spatial sensing assistant equipped with a camera, a circular touchscreen, and voice interaction. It processes vision tasks locally on-device, then connects to large model services to understand scenes and trigger automated actions.
A fingertip-sized development board paired with a detachable camera expansion board and an onboard digital microphone. It is ideal for exploring TinyML image classification, sound recognition, and connected sensing projects.
A low-power edge vision module integrating a neural network accelerator to deploy image classification and object detection models. The camera must be connected separately, and it can be paired with XIAO or other host boards.
A maker-oriented programmable Linux AI camera integrating a camera sensor and edge inference capabilities. It supports orchestrating visual recognition and automation workflows via Node-RED: shown here is the 2002w version with Wi-Fi support.
An edge computing host powered by Jetson Orin NX 16GB, capable of deploying complex vision models and adapted generative AI applications. It is equipped with a cooling enclosure and standard expansion interfaces; cameras and other peripherals must be connected separately.
An all-in-one multimodal AI development device integrating a camera, touchscreen, and audio unit. It enables rapid deployment of vision and voice applications using MaixPy, and can also experiment with running adapted small language models locally.
A portable vision development device with a built-in camera and touchscreen that runs models like object detection locally. It supports writing logic in MaixPy, ideal for quickly prototyping AI projects with direct on-screen previews.
A micro Linux development board powered by the SG2002 chip with a dedicated built-in neural network accelerator. It supports external cameras and screens for building local lightweight visual recognition and edge computing projects.
A smart camera module that integrates image acquisition and neural network inference on the same sensor chip. It must be connected to a Raspberry Pi board to operate, making it ideal for completing local vision tasks like object detection directly on the sensor.
A dedicated expansion board that adds local AI vision inference capability to the Raspberry Pi 5. Available in different compute performance tiers, it handles tasks like object detection and pose estimation, and must be used with a Raspberry Pi 5.
A dedicated AI expansion board for the Raspberry Pi 5 with its own onboard memory, capable of running adapted large language models, vision-language models, and vision tasks locally. It requires a Raspberry Pi 5 and corresponding model software to operate.
An official Jetson developer kit designed for local generative AI and robotics development. It can deploy and run adapted language and vision models on-device; cameras, storage, and other peripherals must be added based on project requirements.
A lightweight depth camera combining color imaging, stereo depth perception, and onboard AI inference into a single device. Running spatial vision models with the DepthAI framework, it provides distance and object information for robotics.
A stereo AI spatial camera featuring a rugged metal enclosure that simultaneously delivers color imagery, depth measurement, and onboard model inference. Connected to a host via USB, it is well-suited for robotic perception and spatial detection scenarios.
A smart vision development board in a miniature form factor, integrating an onboard camera, microphone, motion sensor, and distance sensor. It runs TinyML models on-device for low-power image classification, sound recognition, and environmental sensing.
A low-power sound recognition board powered by a dedicated neural network processor. It can continuously listen for wake words or specific acoustic events, combining with an onboard motion sensor for local event detection.
A dual-core development board capable of running micro machine learning applications like TensorFlow Lite Micro. It is well-suited for industrial sensing and smart control prototypes; vision projects require an external camera or a dedicated Vision Shield.
An out-of-the-box AI vision sensor equipped with a camera and an interactive touchscreen, preloaded with classic recognition algorithms and supporting custom model deployment. It can output results to a microcontroller or connect to large model applications via the MCP protocol.
An introductory AI learning board featuring an onboard display, camera, and audio unit. It supports graphical programming for quickly building visual recognition and voice control projects; conversational large model capabilities require connecting to cloud network services.
A multimedia audio development board equipped with a dual-microphone array, supporting acoustic front-end processing, local voice wake-up, and offline command word recognition. It can connect external displays, cameras, and speakers to build connected voice interaction prototypes.
An evaluation board designed for multimedia interaction, offering camera, display, and audio interfaces. It supports deploying adapted lightweight vision models with ESP-DL; onboard Wi-Fi and Bluetooth connectivity are provided by a companion ESP32-C6.
An OpenMV vision development board equipped with a hardware neural network accelerator, featuring an integrated onboard camera and wireless connectivity. It supports scripting vision logic in MicroPython to run adapted recognition models efficiently on-device.
A bionic quadruped robot dog that supports direct gait control using offline voice commands, with open motion programming and customizable voice triggers. Advanced capabilities such as visual recognition or large model interactions require dedicated add-on modules and corresponding software.
Bee Frontier is a wearable AI memory device that clips to clothing or wraps around your wrist. Dual noise-canceling microphones capture conversations and turn them into summaries, to-dos, and reminders. Rated IP67, it runs for roughly 7 days per charge. A single button starts or stops recording; Alexa integration is available via software as part of a Bee Pro subscription.
Bee Eclipse is a credit-card-sized AI recorder that attaches to the back of your phone via MagSafe. It captures calls and in-person conversations, then automatically generates summaries and action items. A single button with a status indicator keeps operation simple, and the battery lasts more than 21 days. Some advanced features require a Bee Pro subscription.
Qwen AI Glasses N1 bring the Qwen assistant to your face. The 50 MP 1G5P lens shoots 4K video, while a bone-conduction microphone picks up your voice for real-time translation, meeting transcription, and voice-driven task execution. The battery is swappable, with up to 9 hours of music playback per charge.
Qwen AI Glasses N1 Pro builds on the N1 with eye-tracking interaction and iris recognition for identity verification. It also monitors heart rate, blood oxygen, and skin temperature. The 50 MP camera and Qwen assistant carry over from the base model, with a runtime of around 8 hours.
Qwen Clip is an open-ear clip-on earphone co-developed by Alibaba Qwen and Bose. It delivers real-time translation across 89 languages, AI note-taking, and voice-triggered task management. A single earbud runs for more than 9 hours; the charging case extends total playtime to over 40 hours.
PaperDIY is the ESP32-S3-powered successor to the M5Paper. It pairs an ESP32-S3 N16R8 with a 4.7-inch 960×540 ED047TC1 e-ink display that renders 16 levels of gray. An M5PM1 power management module, microSD slot, and 140 mAh battery round out the package. Available as a fully assembled unit or as an assembly kit.
T5 e-Paper Basic combines an ESP32-S3 with a 4.7-inch 960×540 e-paper display. An AXP2602 fuel gauge tracks battery level while an XL9555 IO expander adds extra GPIO. Five physical buttons and 16 MB of flash with 8 MB of RAM make it a capable platform for info displays and electronic name badges.
The XIAO IPS Touch Display Series spans six boards featuring either an ESP32-S3 Plus or nRF52840 Plus, with screen sizes of 0.96, 1.14, and 1.47 inches. Touch input is exclusive to the 1.47-inch variant; Wi-Fi is available only on ESP32-S3 models. All versions include an IMU, microphone, and battery connector.
Orange Pi O1 is a single-board computer powered by the SOPHGO BM1688, whose built-in TPU/NPU delivers 16 TOPS of INT8 inference. Eight Arm Cortex-A53 cores handle general processing. Choose from 8 GB or 16 GB of RAM. Connectivity covers M.2 NVMe storage, dual Gigabit Ethernet, WiFi 6E, two MIPI CSI camera ports, and optional 4G or 5G, making it well suited for multi-channel video analytics.
UGen300 and UGen200 are plug-and-play Hailo edge AI accelerators from ASUS. The UGen300 uses a Hailo-10H chip and comes in two form factors: a USB dongle with 8 GB of on-board memory, and an M.2 module available with 2 GB or 8 GB. The UGen200 uses a Hailo-8L chip and is offered as an M.2 module only.
Coralboard SL and SR are low-power edge AI development platforms built jointly by Synaptics and Google Research. The SL variant is driven by the Astra SL2619 and integrates a Coral NPU. The Torq compiler toolchain is included to streamline model deployment.
VIEWE's 4.3-inch ESP32-S31 Display Dev Board is among the first third-party boards to adopt the ESP32-S31. It pairs the new chip with a 4.3-inch 800×480 display, offering an early entry point for exploring the ESP32-S31 platform.
WT32P4C5-43S is a 4.3-inch touch display development board combining an ESP32-P4 main processor with an ESP32-C5 for dual-band WiFi 6 connectivity. The two-chip design targets applications that need strong processing alongside next-generation wireless performance.
ToughC5 is a weatherproof IoT controller built around the ESP32-C5. Its 2-inch high-brightness screen stays readable outdoors, and the radio stack covers dual-band WiFi 6, BLE 5, and Zigbee/Thread, making it a versatile hub for outdoor smart deployments.
Romu is a tiny 15.5×12 mm development board built around the RP2354A microcontroller. It connects via USB 2.0 and is currently available through a crowdfunding campaign.
PlugMate and PlugClaw are thumb-sized personal computers running PlugOS, an Android-based OS. Plug either into a phone or PC's USB-C port and they borrow the host's display and network connection. PlugClaw adds support for running OpenClaw inside an Ubuntu virtual machine.
Pixl Nub is a desktop robot that senses presence and focus using dual radars — no camera required. Presence sensing runs local-first and works offline. It tracks focus duration, interruptions, and the time needed to return to focus. Power and data run over a single USB-C connection.
Notra Neo is a MagSafe-mounted AI meeting recorder from BOYA. With 8 GB of built-in storage, it transcribes and summarizes meetings in more than 140 languages. Local recording works without a network connection.
JARVIS Desk Bot is an open-source talking desktop robot powered by the ESP32-S3. It serves as a voice front end for the Hermes Agent and supports locally hosted Xiaozhi, Parakeet, and Kokoro speech services. Wake it with the word 'Jarvis'.
Waveguide AI glasses with a dual-chip design and a stated weight of about 45g. Replaceable frames and nose pads support different fits, while the product focuses on voice assistance, immediate visual interaction and everyday information prompts.
AR display glasses for entertainment and gaming, offering a 160-inch-equivalent virtual screen, 2D/3D switching and adjustable myopia correction from 0 to 6 diopters. They work as a wearable external display for a portable large-screen experience.
A touchscreen scanning translator pen that turns printed text into digital text and reads it aloud. Voice and photo translation, dictionary lookup and reading assistance support language study and text capture. Language availability depends on the software version.
A control module that recognizes preset voice commands locally without internet access, with support for voice self-learning to record new commands. It transmits results to a host controller via I2C or UART, ideal for voice-controlled lighting and interactive robots.
A miniature two-wheel educational robot with proximity, light, infrared and three-axis acceleration sensing. Scratch, Python and other coding environments support hands-on exercises in line following, obstacle avoidance, mazes and motion control.
AI glasses weighing about 49g, combining binocular Micro LED waveguide displays, a 12MP camera and Snapdragon AR1. Voice assistance, live translation captions, teleprompting and navigation bring information into view while supporting hands-free first-person capture.
An OpenMV camera development board powered by a dual-core processor and dual neural network accelerators. It enables building low-power vision recognition applications with MicroPython; wireless connectivity requires an external expansion module.
A 4K PTZ webcam with a 1/1.5-inch CMOS sensor. AI tracking and motorized pan-and-tilt movement keep the subject in frame, with USB connectivity for video meetings, teaching and desktop streaming.
A 360 action camera with dual 1/1.28-inch sensors and a triple AI chip imaging system. It offers 8K 30fps spherical video, replaceable lenses and FlowState stabilization, allowing the framing direction to be chosen after capture.
An open-source LoRa pager from exploitee.rs, built around ESP32-S3 and SX1262. Send and read Meshtastic messages directly on the device without a phone, cellular network or Internet. A green-backlit LCD, packet logging, spectrum tools and CHIP-8 games make it a portable radio development tool. The official store lists it from US$250; sold out when checked.
Memoket Gem is an 11 g wearable recorder with dual microphones, worn on a wristband or an Apple Watch co-wear band. Audio is stored on the device, transferred to the app over Bluetooth or Wi-Fi, then transcribed and summarized by cloud services into notes, action items and cross-conversation context. The manufacturer specifies up to 20 hours of continuous recording; the reference hardware price is US$169.