With the rapid development of artificial intelligence technology, various industries are actively exploring the application of AI technology to achieve intelligent transformation. In the consumer electronics market, AI technology has become a key factor driving product innovation and market growth. The application of AI technology not only enhances the functionality of products but also provides users with a more convenient and personalized operation and usage experience.
In the home appliance sector, AI technology also offers many possibilities for product development. NXP has been deeply involved in the home appliance field, with numerous successful MCU cases in home appliance control boards, motor drives, display screens, touch buttons, and more. In today's fast-evolving AI technology landscape, NXP has kept pace with the times by launching MCUs with NPU functionality that can accelerate AI algorithms. The technical team at the World Peace Group (WPG) has developed the "AI Coffee Capsule Recognition Solution based on NXP MCX N947 MCU through NPU" to provide customers with reference ideas for AI product directions and offer initial development support to assist customers in faster AI product development.
This solution fully utilizes the powerful performance of the NXP MCX N947 high-performance microcontroller (MCU) integrated with a neural processing unit (NPU), aiming to develop a coffee machine with true intelligent recognition capabilities for AI coffee capsule recognition. By leveraging the neural network algorithm acceleration capabilities of the NPU, this solution can achieve rapid and accurate recognition of coffee capsules and automatically adjust brewing parameters based on user preferences, providing a personalized coffee experience.
In the following content, we will elaborate on the main control introduction, solution block diagram, function introduction, and demo demonstration.
1. Main Control of the Solution
The main control of this solution is the new series of NXP MCX N series products, which are high-performance, low-power microcontrollers equipped with intelligent peripherals and accelerators, providing multitasking capabilities and high energy efficiency. The selected MCU from the MCX series is the N947, which integrates an NPU and supports eIQ tool development.
Below are the detailed specifications of the main control chip:
- Core: Dual-core Cortex M33
- Clock Frequency: 150MHz
- Integrated Functions: DSP, NPU
- Peripherals: UART, USB, CAN, FlexIO, FlexPWM, FlexSPI
- Operating Voltage: 1.71V to 3.6V
- Operating Temperature: -40 to 105 °C
- Package: VFBGA184: 9 x 9 x 0.86 mm, 0.5 mm pitch, HLQFP100: 14 x 14 x 1.4 mm, 0.5 mm pitch
The efficient NPU features of the MCX N947 MCU and the convenient eIQ tool development make it the best choice for implementing our AI coffee capsule recognition solution.
2. Solution Block Diagram
To accelerate the development process, this solution directly uses the FRDM board of the MCX N947 for demo setup, requiring an additional camera for image data collection and a display screen to show recognition results. These three components are sufficient to implement this solution. The camera used is the OV7670, with a resolution of 640*480, and the screen is paired with the NXP LCD-PAR-S035, 3.5 inches, with a resolution of 480x320, which can be applied to various NXP FRDM boards with a common interface.
3. Function Introduction
(1) Utilizing the NPU core and eIQ Neutron software library, providing ML acceleration to achieve recognition of 8 types of coffee capsules
(Americano, Americano_Intenso, Americano_Rich_Aroma, Doppio_Espresso, Espresso_Napoli, Grande, New_York_Morning_Blend, Palermo)
(2) Using the built-in model mobilenet v1 (Flash: 225K, RAM: 144K) in eIQ Portal for model inference
(3) The camera captures images with a resolution of 384*384
(4) The camera image data is internally converted to a resolution of 128*128 before being fed to the model for inference
(5) Supports button control to start recognition, capturing images from the camera within 1 second (configurable), and displaying the recognized coffee type on the LCD after completion
By capturing images of coffee capsules with the camera, the type of coffee capsule is recognized and displayed on the screen along with brewing suggestions.
The above is an introduction to the AI coffee capsule recognition solution. There is still room for optimization, and for specific product applications, more user-friendly features can be added, such as more attractive interface displays and support for touch buttons, all of which can be developed using the powerful MCX N947. This MCU is not limited to a single application for developing AI functions; there are many more application directions, and we welcome everyone to discuss together.
►Scene Application Diagram
QianTu Network - QianTu Network
►Display Board Photos
►Solution Block Diagram
►Core Technical Advantages
1. High performance - Dual-core Cortex M33 with a high frequency of 150MHz
2. AI acceleration function - Integrated NPU
3. Rich peripherals - USB, CAN, FlexIO for simulating camera and screen interfaces
4. Convenient development - Supports eIQ tool development for NPU
►Solution Specifications
1. Supports camera image capture with a resolution of 640*480
2. Supports screen display with a resolution of 480*320
3. Supports recognition of 8 types of coffee capsules (can also learn and train additional coffee types through eIQ)
4. Supports rapid recognition (recognition occurs within 1 second after pressing the recognition button)