ESP-SparkBot is a compact AI-powered robot built on the ESP32-S3, offering voice interaction, image recognition, and multimedia capabilities. Despite its small size, it integrates a wide range of features, including an AI assistant for conversations, weather updates, and music playback. Users can interact with it through touch sensors or voice commands, allowing for a seamless experience. The robot also includes a foldable camera for face recognition and motion detection, enabling a variety of smart applications. It supports over 200 offline voice commands and can connect to large language models like ChatGPT, Wenxin Yiyan, Doubao, DeepSeek, Kimi, Zhipu AI, Mistral AI and Xiao zhi AI for more advanced AI-driven interactions.
Beyond being a virtual assistant, ESP-SparkBot can function as a mini entertainment system. It features USB screen mirroring, allowing users to stream videos and play games without requiring an HDMI connection. The device also supports wireless control via a browser-based interface, eliminating the need for a dedicated app. Additionally, it includes an accelerometer for gesture-based interactions, making it suitable for games like 2048 or as a digital dice that responds to physical movement. The ESP-SparkBot is capable of synchronising multiple units via ESP-NOW.
In terms of expandability, ESP-SparkBot’s modular design supports additional attachments, including a magnetic charging port and a track-based chassis for remote-controlled movement. This allows the robot to function as a mobile surveillance unit, streaming video in real-time while being controlled wirelessly. Its two-way audio capability also lets it act as a microphone and speaker. Bluetooth connectivity via the Xiaodu mobile app enables smart home integration, letting users control devices or switch between different voice profiles. The robot’s versatile applications make it an engaging tool for AI experimentation and everyday use.
Designed for developers and AI enthusiasts, ESP-SparkBot is completely open-source, with both hardware and software resources available for customisation and links to them can be found in the ESP-Spark project page.
