astroai am33d how to use

AstroAI AM33D: A Step-by-Step Guide to Using This Powerful Scope

By Anthony Moore
AD

Embarking on your journey to understanding the AM335x System-on-Chip (SoC) by TI with the AstroAI platform involves a fascinating world of hardware-as-a-service and AI-on-the-edge. In this comprehensive guide, we'll delve into the intricacies of the AM335x series,-zeroing in on the AM3356 variant, and guide you through the process of using it effectively with AstroAI.

To fully grasp the potential of the AM3356 with AstroAI, let's first understand the power and flexibility that this SoC brings to the table. Powered by a single-core ARM Cortex-A8 cores running at 720MHz and accompanied by a NEON media processing engine, the AM3356 offers a potent combination of performance and power efficiency. Moreover, it boasts integrated peripherals like USB, CAN, I2C, and UART, making it an excellent choice for a multitude of IoT and edge computing applications.

AstroAI Platform and AM3356

AstroAI is a powerful platform developed by CEVA to accelerate AI at the edge. By leveraging the AM3356's capabilities, AstroAI enables developers to run complex AI tasks directly on the AM3356 without relying on the cloud, ensuring real-time processing and preserving data privacy.

AD
Make your own astrolabe - 1. Introduction
Make your own astrolabe - 1. Introduction
How to Take Flat Frames for Astrophotography [The Easy Way]
How to Take Flat Frames for Astrophotography [The Easy Way]
Making an Astronaut Using Paper and Tape
Making an Astronaut Using Paper and Tape

To commence your exploration, you'll first need to set up your development environment. This involves installing the required software, including the AstroAI SDK, the GNU MCU tools for AM335x, and OpenOCD for flashing and debugging your device. Once set up, you'll be ready to dive into the fascinating world of AI-on-the-edge with the AM3356.

Setting Up Your AM3356 Board for AstroAI

Hardware Preparation: Before you begin, ensure your AM3356 board is properly connected to your host computer via USB. Additionally, if your board has a JTAG interface, connect it to your debugger/trace pod. This will enable you to flash your board and monitor its functionality during development.

a hand is holding a small toy in front of a blue and white box with an astronaut on it
a hand is holding a small toy in front of a blue and white box with an astronaut on it

Software Installation and Configuration: Now, let's turn our attention to the software setup. Begin by installing the GNU MCU tools for AM335x, which provide an ARM toolchain for building and debugging applications. Next, install OpenOCD, a platform for on-chip debugging and testing. Finally, install the AstroAI SDK, which contains the necessary frameworks and models for running AI tasks on your AM3356 board.

Compiling and Flashing Your First AstroAI Model

Now that your environment is configured, let's compile and flash a simple AI model onto your AM3356 board. For this example, we'll use a model from the AstroAI sample repository. Choose a simple model, such as 'mobilenet_v1_1.0_224_quant.tflite', a quantized version of the MobileNetV1 model for image classification.

AD

Step-by-step process:

  1. Navigate to your AstroAI SDK installation directory and clone the sample models repository: git clone https://github.com/ceva-dsp/astroai-samples.git
  2. Navigate to the cloned repository: cd astroai-samples/models
  3. Copy the chosen model file to your local development directory
  4. Create a new C project using the AstroAI Makefile template: make init mismatched-boards/am335x EVB am335x\defconfigs\am335x_evm.h TARGET_METADATA=tflite_model_demo/mobilenet_v1_1.0_224_quant.tflite
  5. Compile the project: make
  6. Flash the compiled binary onto your AM3356 board using OpenOCD: make flash
After a successful flash, your AM3356 board will begin running the AI model and is now ready to process AI tasks at the edge.

Exploring Advanced Features of AstroAI and AM3356

With the basics covered, let's delve into the more advanced features that the AstroAI platform and AM33356 SoC offer. AstroAI supports a wide range of AI frameworks, including TensorFlow, PyTorch, and ONNX, enabling you to use pre-trained models or develop custom ones directly in your preferred framework.

AD

Moreover, the AM3356's NEON media processing engine allows for accelerated performance of AI tasks. To take full advantage of this, you can utilize the AstroAI SDK's optimizations and Preparatory Neural Network Compiler (PNNC) tool to generate custom binary kernels for your models, further enhancing performance.

Deploying Custom AI Models with AstroAI

To deploy your custom AI models on the AM3356 using AstroAI, follow these steps:

  1. Train and export your model using your preferred AI framework
  2. Quantize your model using tools like TensorFlow Lite's quantization tools or the AstroAI SDK's built-in quantizer
  3. Compile your quantized model using the AstroAI PNNC tool to generate optimized binary kernels for the AM3356
  4. Integrate your custom AI model into your application using the AstroAI SDK's APIs
  5. Compile and flash your application to your AM3356 board, following the steps outlined in the previous section
By following these steps, you can seamlessly integrate your custom AI models into your applications, using the AstroAI platform and AM3356 SoC to process complex AI tasks efficiently at the edge.

AD

Monitoring and Debugging Your AM3356 Applications

Efficiently monitoring and debugging your AM3356 applications is crucial for identifying and resolving issues quickly. AstroAI SDK provides tools like the AstroAI Logger for printing debug messages directly from your application. Additionally, you can leverage the GNU MCU tools and OpenOCD for debugging your application using a JTAG interface.

To enable debugging, include necessary lines in your Makefile, such as DEBUG Enabled and DEBUG_LOG Enabled, and connect your debugger/trace pod to the AM3356 board's JTAG interface. Once configured, you can use tools like GDB to step through your code and inspect variables for a more in-depth understanding of your application's behavior.

Finally, as you venture deeper into the realm of AI-on-the-edge with AstroAI and AM3356, the possibilities are endless. Fromacija IoT devices to enhancing industrial automation, or even developing innovative AI-enabled consumer products, your creativity is the only limit. Happy exploring, and may your journey with AstroAI and AM3356 be as enlightening as it is exciting!

AD