ace-trump-tech/MindPaw

What it solves

MindPaw provides an affordable, open-source platform for building a desktop quadruped robot. It bridges the gap between simple remote-controlled toys and complex AI robots by integrating voice interaction, gesture recognition, and LLM-powered conversation into a low-cost hardware package (approximately ¥50).

How it works

Built on the ESP8266 microcontroller, the robot uses a combination of specialized modules and software logic:

• AI Integration: Connects to the Doubao LLM API for intelligent conversations, responding with a mix of text, expressions, and physical movements.
• Perception: Uses an OV2640 camera for gesture recognition (via a lightweight MLP model) and an HLK-V20 module for voice command recognition.
• Interaction: An OLED screen displays 7 different emotions, while a speaker provides audio feedback. A web-based interface allows for manual remote control via a browser.
• Emotion Engine: Implements a PAD model state machine to simulate emotional responses based on interaction frequency and content.

Who it’s for

• Hobbyists and students wanting to learn how to integrate microcontrollers, servos, cameras, and AI APIs.
• Makers looking for a budget-friendly entry point into robotics and embodied AI.
• Developers interested in implementing lightweight neural networks on embedded hardware.

Highlights

• Extreme Low Cost: Total material cost is roughly ‚50 (basic version ‚35).
• Multimodal Interaction: Combines voice, gestures, and web controls to trigger actions.
• Embodied AI: Uses an LLM to drive physical reactions and emotional expressions.
• Fully Open Source: Includes all C++ source code, PCB designs (LCSC/JLCPCB), and 3D printable STL files.