I have been busy researching lifestyle action cameras to capture all my outdoor adventures. Naturally, I looked to GoPro, but also wondered if there was something else. In my research, I came across a mini cube camera that looks perfect for capturing my outdoor activities. Now, there’s another cube camera aimed at AI inferencing ready to capture all your edge application developments.
Artificial intelligence (AI) requires extreme computational horsepower. “Maxim Integrated®, now a part of Analog Devices, is leveraging its experience as a leading supplier of ultra-low power microcontrollers to eliminate the energy spent on edge AI computations. This blog will discuss how the MAXREFDES178 Cube Camera reference design is cutting the power cords long associated with achieving AI insights.”
The promise of machine vision (MV) and other advanced AI inferences running in a battery-powered device is compelling. Analog Devices noticed three key areas of AI inferencing that they could provide a solution for. These areas included delivering a solution that could:
To help address all these concerns, Analog Devices developed the AI Cube Camera, also known as the MAXREFDES178 (Figure 1).
Figure 1: The MAXREFDES178 AI Cube Camera addresses three critical issues for developers: demonstrating use cases, employing low-power levels, and lowering the resources barrier to entry. (Source: Maxim Integrated, now part of Analog Devices)
The MAXREFDES178 is a cube camera reference design based on the MAX78000 and MAX32666 microcontrollers (MCUs). It helps AI at the edge device designers accelerate their proof-of-concept (PoC) to the market phase for AI applications, including robotics. The MAX78000 is a unique AI microcontroller built to enable neural networks (NN) to execute at ultra-low power and live at the edge of the IoT. Analog Devices’ hardware-based convolutional neural network (CNN) accelerator enables battery-powered applications to execute machine vision and other AI inferences while spending only microjoules (µJ) of energy. The unique energy-efficient architecture of the MAX78000 enables battery-powered AI at the edge applications such as face identification, large vocabulary, keyword spotting, scene segmentation, and object detection.
While the MAXREFDES178 cube camera is small, there is a lot of functionality packed into it: a camera, microphone, SD card slot, touch screen, buttons, and LEDs provide a way to interact with the real world. On the inside, there is a lot of silicon (Si) content: a Bluetooth® microcontroller, a battery charger, audio codec, and two MAX78000 AI processors—one is connected to an image sensor and is intended for image and video applications, while the other MAX78000 is linked to the microphone input and is designed for voice and audio applications. Even with all that, the cube camera has a fair amount of open space inside for future expansion (Figure 2).
Figure 2: An internal picture of the AI Cube Camera shows sufficient room for future expansion while it is jammed packed with features. (Source: Maxim Integrated, now part of Analog Devices)
An Android app is available on the Google Play store to help manage the cube camera. For example, along with the Face ID demo, you can take a picture of your face, and a new embedded image will be sent to the cube camera. The Face ID demo will start to recognize you, not just the celebrities from the default database.
The AI Cube Camera can be used for a variety of applications. This includes audio processing involving voice activation control/word recognition/multi-keyword recognition, sound classification, or noise cancellation (Figure 3).
Figure 3: The AI Cube Camera can perform audio processing because it has a digital microphone, multiple audio codecs with stereo audio input and output. (Source: Mouser Electronics)
It can also perform facial recognition and identification through picture analysis, object detection, and classification (Figure 4).
Figure 4: The AI Cube Camera is capable of processing images through a trained convolutional neural network (CNN). (Source: Mouser Electronics)
Additionally, it may perform time-series data analysis for such things as heart rate/health signal analysis, multi-sensor analysis, and predictive maintenance (Figure 5).
Figure 5: The AI Cube Camera may be employed for predictive maintenance (PdM) because it is capable of time-series data analysis. (Source: Mouser Electronics)
The MAXREFDES178 Cube Camera, powered by the MAX78000 AI processor, is a real and tangible reference design. Not only does it confirm that great engineering minds often think alike, in this case, it squares. The Cube Camera will help designers spark their imagination, proves that edge AI is ready to go, and even helps you convince your boss to do something extraordinary for your next product. Let your engineering journey begin.
Paul Golata joined Mouser Electronics in 2011. As a Senior Technology Specialist, Paul contributes to Mouser’s success through driving strategic leadership, tactical execution, and the overall product-line and marketing directions for advanced technology related products. He provides design engineers with the latest information and trends in electrical engineering by delivering unique and valuable technical content that facilitates and enhances Mouser Electronics as the preferred distributor of choice.
Before joining Mouser Electronics, Paul served in various manufacturing, marketing, and sales related roles for Hughes Aircraft Company, Melles Griot, Piper Jaffray, Balzers Optics, JDSU, and Arrow Electronics. He holds a BSEET from the DeVry Institute of Technology (Chicago, IL); an MBA from Pepperdine University (Malibu, CA); an MDiv w/BL from Southwestern Baptist Theological Seminary (Fort Worth, TX); and a PhD from Southwestern Baptist Theological Seminary (Fort Worth, TX).
