Compact AI Vision Sensor

November 11, 2025

The HUSKYLENS 2 is a compact AI vision sensor powered by a 6 TOPS dual-core processor, capable of running custom YOLO models and delivering real-time contextual perception for robotics, education, and intelligent systems. The HUSKYLENS 2 K230 is a next-generation AI camera designed to bring powerful machine vision to embedded systems and edge devices. Built around the Kendryte K230 dual-core processor running at 1.6GHz, it features an integrated 6 TOPS neural accelerator that enables high-speed, on-device inference for complex visual tasks—without relying on cloud computation.

One of its standout capabilities is support for over 20 built-in AI models, including face recognition, object tracking, line following, and OCR. These models are optimized for low-latency execution and can be deployed instantly, making HUSKYLENS 2 ideal for rapid prototyping and educational robotics. For advanced users, the device supports custom model deployment, including YOLO-based architectures.

HUSKYLENS 2 introduces a novel MCP (Multimodal Contextual Perception) service, which allows it to deliver structured, intelligent briefs to large language models. Instead of raw image data, it can inform an LLM that a known person is waving or a red object is moving left, enabling smarter decision-making in AI systems that combine vision and language.

The hardware includes a 2MP GC2093 sensor capable of 60 FPS capture, a 2.4-inch IPS touchscreen with full lamination, a microphone, speaker, RGB indicator, and dual fill lights. Real-time video streaming is available via USB-C or an optional Wi-Fi 6 module. Connectivity is broad, with UART and I2C interfaces for integration with Arduino, Raspberry Pi, ESP32, micro:bit, and UNIHIKER boards. The device runs on 3.3–5V and consumes just 1.5–3W. It includes 1GB LPDDR4 RAM, 8GB eMMC storage, and a TF card slot for expansion.

The video above is intended for educational purposes and does not constitute or imply an endorsement by FutureForAll.org.