DFRobot
Gravity: HUSKYLENS 2 - 6 TOPS LLM MCP AI Vision Sensor (20+ Models, Deploy Custom Models)
· MPN: SEN0638
HUSKYLENS 2 is a powerful AI vision sensor with a 1.6GHz dual-core processor and 6 TOPS AI accelerator. With 20+ built-in AI models and the ability to train ...
Get notified when back in stock
HUSKYLENS 2 is a powerful AI vision sensor with a 1.6GHz dual-core processor and 6 TOPS AI accelerator. With 20+ built-in AI models and the ability to train and deploy custom YOLO models, it brings advanced computer vision to Arduino, Raspberry Pi, and micro:bit projects.
The unique MCP (Model Control Protocol) service provides contextual awareness to Large Language Models, enabling smarter AI decision-making. Run multiple models simultaneously for interactive, intelligent applications.
Key Features
- 6 TOPS AI Accelerator – High-speed, low-latency vision processing
- 20+ Built-in Models – Face detection, object tracking, QR codes, and more
- Custom Model Training – Deploy your own YOLO models
- LLM Integration – MCP service for contextual AI awareness
- Real-Time Video Streaming – Live feed over WiFi
- Replaceable Camera – 2MP module, field-swappable
- Multi-Controller Support – Arduino, Raspberry Pi, micro:bit compatible
Ideal For
- Educational robotics and AI learning
- Object detection and tracking projects
- Interactive installations and smart devices
- Prototyping intelligent vision systems
Specifications
- Processor – 1.6GHz dual-core
- AI Performance – 6 TOPS
- Camera – 2MP (replaceable module)
- Built-in Models – 20+
- Connectivity – UART, I2C, USB
Can be added to the Maqueen Plus V3 as pictured

Jargon buster
Plain-language definitions for the technical terms used above.
- I2C
- I2C is a two-wire communication bus used by many sensors and small modules. It matters because several I2C devices can share the same two wires, but each device needs a compatible address and your controller must support I2C.
- TOPS
- TOPS means trillions of operations per second, often used to describe AI accelerator performance. It helps compare whether a computing module is suited to lightweight image recognition or more demanding neural-network workloads.
- UART
- UART is a simple asynchronous serial interface that sends data over separate transmit and receive wires, usually labelled TX and RX, with both ends set to the same baud rate. It is a common way for microcontrollers and other serial devices to exchange data.
Find this product in
Brands
Sensors & Input
STEM & Education
FCC Certificate
Compliance · 151.3 KB · Click any page to view full size
RoHS Certificate
Compliance · 714.0 KB · Click any page to view full size
EN71 Certificate
Compliance · 151.6 KB · Click any page to view full size
Supplier page — dfrobot.com
Supplier Description · 1.1 MB · Click any page to view full size
Resources & Downloads
Guides, code examples, and more
Source Code
Open-source libraries, firmware & example projects for this product
659aaf6
2 months ago
· 71 commits
- examples add donkey car demo 6 months ago
- python Add files via upload 2 months ago
- DFRobot_HuskylensV2.cpp fix: In low-memory mode, increase the send buffer to 64 bytes. 6 months ago
- DFRobot_HuskylensV2.h 放弃colorID,使用rgb格式的color 8 months ago
- HuskyLens2_Protocol.md Update protocol documentation for total_blocks fields 6 months ago
- HuskyLens2_Protocol_CN.md Update data table in HuskyLens2 protocol document 6 months ago
- keywords.txt modify protocol 7 months ago
- library.properties update library.properties 6 months ago
- LICENSE.txt first commmit 8 months ago
- ProtocolV2.cpp Fix: chunk I2C writes to prevent incomplete transfer on low-mem devices 5 months ago
- ProtocolV2.h chang getResult(eAlgorithm_t algo) to getResult(eAlgorithm_t &algo) 6 months ago
- README.md Simplify setMultiAlgorithm by removing extra params 7 months ago
- README_CN.md chang getResult(eAlgorithm_t algo) to getResult(eAlgorithm_t &algo) 6 months ago
- Result.cpp Remove redundant code 6 months ago
- Result.h Merge branch 'master' of github.com:DFRobot/DFRobot_HuskylensV2 5 months ago
Related Tutorials
Free guides on learn.littlebird.com.au