UNIHIKER K10 AI Agent Coding Board for STEM & Maker (Edge LLM, Vision AI, Voice Chatbot, ESP-CLAW Support)

The UNIHIKER K10 is an ESP32-S3 powered AI coding board designed for STEM students, educators, and makers exploring physical AI. Equipped with a 2.8-inch colour display, an onboard 2MP camera, environmental sensors, and dual wireless connectivity, it bridges cloud LLMs and edge computing. It provides an accessible platform for vision AI, offline voice control, and IoT experimentation without complex wiring.

Pre-loaded vision AI models

The integrated 2MP camera pairs with four pre-loaded vision models covering face detection, cat and dog recognition, QR code scanning, and motion sensing. Beginners can build smart monitoring and interactive visual projects straight out of the box without needing complex setup.

Offline speech recognition and control

An onboard microphone and pre-installed voice models enable local speech processing without requiring an active internet connection. Users can program custom voice commands to trigger physical outputs and create standalone interactive systems.

Custom TinyML model deployment

Beyond the pre-installed tools, the board supports TinyML for training and deploying custom vision and sensor models. This lets makers collect bespoke datasets and run personalised machine learning models directly on the device.

Real-time colour data display

The integrated 2.8-inch 240x320 colour screen provides immediate visual feedback for sensor readings, AI outputs, and user interfaces. This eliminates the need for external monitors during testing and portable deployments.

Onboard sensors and wireless IoT

Equipped with temperature, humidity, light, and accelerometer sensors alongside 2.4 GHz Wi-Fi and Bluetooth 5.0, the board is suited for continuous IoT data logging. It can monitor environmental conditions locally or transmit telemetry to remote dashboards.

Versatile I/O and expansion ports

Hardware interfaces include USB Type-C, a MicroSD card slot, Gravity 3-pin and 4-pin ports, edge connectors, and a 2-pin PH2.0 battery terminal. These ports allow quick connection of external actuators, sensors, and portable power sources.

Flexible coding and ESP-CLAW support

The board supports beginner-friendly drag-and-drop block coding through Mind+ as well as MicroPython for advanced development. It is also an officially supported platform for the ESP-CLAW framework, enabling natural-language hardware control via chat applications and cloud LLMs.

Video

Downloads & documentation

• Product wiki (DFRobot)