PSOC™ Developer Journey

This new page will be updated regularly based on feedback, with additional development flows and content coming soon. Check back for updates.

Develop

Choose from one of the development workflows by clicking on it to open the steps.

The MCU development flow is for developers who wish to use ModusToolbox™ to develop their MCU based application.

1. Download ModusToolbox™ setup

Use ModusToolbox™ Setup tool for simplest installation as there are a set of tools to be included. The Setup tool allows you to select the tools needed and they will be installed together.

Open resource
Open in media player

2. Learn about ModusToolbox™

ModusToolbox™ is a rich ecosystem of tools for MCU development. More information and training videos are available for ModusToolbox™ in the resource button below.

Open resource
Open in media player

3. Getting started with ModusToolbox™

Basic steps to create, build, and debug a hello world application.

Open resource
Open in media player

4. Set up your hardware using Device Configurator

To adapt an existing development board to add sensors or change IO configurations you will use Device Configurator. Device Configurator is included in ModusToolbox™. It presents a graphical view of the device peripherals to allow you to setup, modify and configure the chip. It will then generate macros, data structures and initialization functions for your project.

A complete user guide for Device Configurator is included in the Resource button below.

Open resource
Open in media player

5. Setup your custom hardware using BSP Assistant

If you are starting with an empty project and building up your application for your hardware you should start with BSP Assistant. BSP Assistant is included with ModusToolbox™.

The BSP Assistant helps you create custom Board Support Packages (BSP) for ModusToolbox™ applications. All ModusToolbox™ Applications require a target BSP. Infineon provides BSPs for all our kits as well as for any chip architecture to use as a starting point.

BSP Assistant is included in ModusToolbox™. A complete guide to BSP assistant is included in the Resource button below.

Open resource

6. Add application functionality using Library Manager

The Library Manager is included in ModusToolbox™ and provides a GUI tool for adding or removing middleware libraries within your project. Using the library manager ensures other files or dependencies are included in the project.

The libraries provided are grouped in categories to make selection easier. Select capabilities such as wireless connectivity, graphics, peripheral drivers, sensors and other middleware for your project.

Open resource

7. Build your application

The build system is based on GNU Make. It performs application builds and provides the logic to launch tools and run utilities. Each application has a set of makefiles including a start.mk to setup the environment and bring in the appropriate libraries, and a BSP.mk to bring in required BSP functionality. The top level application makefile sets basic and advanced configuration options and paths. These and additional make files form the build system.

Open resource

8. Debugging your application

ModusToolbox™ supports using OpenOCB using a GBD server and supports the J-Link debug probe.

Various IDEs are supported for establishing debug sessions including Eclipse, VS Code, IAR, and Keil uVision.

See section 3.5 of the ModusToolbox™ User's Guide in the Resource button below.

Open resource

DEEPCRAFT™ Studio: An end-to-end platform for ML on edge devices. Streamlines the entire workflow from data collection and annotation to model deployment. Handle all stages efficiently—data management, model building, evaluation, and edge deployment—in one comprehensive solution.

1. Download DEEPCRAFT Studio

DEEPCRAFT Studio is a development environment for building ML models. You can build and train your model, manage your data and deploy your model to your device all from one tool.

Open resource
Open in media player

2. Model Selection Options

One of the choice to be made as you start your AI journey is on the model itself. DEEPSCRAFT Studio supports building your own model and that is detailed in the steps in this journey. You can also shorten your time to develop with other options on models:

- Bring your own model - if there are 3rd party models suitable for your application, these can be imported and optimized.

- Ready Models - there are a series of ready models available for audio, capacitive sensing, radar and IMU use cases that are production ready to include in your application

- Starter Models - there is a series of started models that include datasets, preprocessing, model architecture and instructions that developers can use to develop your own production ready model.

DEEPCRAFT Studio supports and provides these models to shorten your developer journey, but you can continue this flow to build your own model from scratch.

Open resource

3. Data Preparation

Data Preparation is the initial phase where data is gathered from various sources, such as databases, sensors, or manual collection. The objective is to assemble a comprehensive dataset that accurately represents the problem you aim to solve.

Data collection is done either by importing data into DEEPCRAFT Studio, or doing real time data collection from sensors or any development board. Infineon offers a streaming application integrated with DEEPCRAFT Studio to simplify real time data collection on our development boards. Once data is collected it is labeled to assist the machine learning model in making predictions. Labeling can be done manually, ML assisted or sequentially.

After data is collected and labeled, it is distributed into different sets such as training, validation and test sets to enhance the model is trained, validated and tested on different subsets of data for best performance.

Open resource
Open in media player

4. Preprocessing

After distributing the data into different data sets, the next step is to configure the preprocessor and generate a preprocessor track.

In a regression project, for one-dimensional data, the minimum preprocessing required is applying a sliding window. However, for data more than one-dimensional, a sliding window is not necessary for preprocessing.

Open resource
Open in media player

5. Model Training

After you collect the data and design the preprocessor, the data is passed to the model for training. Before the model can be trained, you need to generate the model and define the layers of the model.

Studio allows generation of multiple different models to be trained and compared to find the best fitting model. DEEPCRAFT Studio can generate a model using the built in Model Wizard, creating a model from scratch using an Empty Model or importing an existing Keras H5 model from another source.

Open resource
Open in media player

6. Model Evaluation

Model evaluation is the process of using multiple statistics and metrics to analyze the performance of a trained model, highlighting both its strengths and weaknesses. Studio provides various methods to evaluate the classification and regression models, such confusion matrix, window visualization, evaluation using Grad-CAM, R-squared (Coefficient of Determination), Mean Squared Error (MSE), Mean Absolute Error (MAE), Root Mean Squared Error (RMSE) and graphical plots tools such as Quantile - Quantile and Histogram of residuals.

Additionally, you can use Graph UX to evaluate the model performance. Graph UX supports real-time model evaluation functionality which helps in analyzing and monitoring the model predictions before deploying a model to production. It also ensures that the model generates accurate predictions on real-time data.

Open resource
Open in media player

7. Model Deployment

After generating the code, the final step involves deploying the model onto the board. This is done by incorporating the model into a deployment application for the MCU targeted. Infineon provides a number of sample deployment applications and you should select one closest to your targeted application and modify it to fit your needs. This is done using the MCU Development flow also references in this Developer Journey guide.

This link below provides detailed instructions on deploying the model on both supported Infineon boards and other development boards. Additionally, this section covers topics such as the Edge API, supported layers and functions, and creating custom layers or functions for advanced users.

Open resource
Open in media player

Use DEEPCRAFT Voice Assistant to develop your voice related application including custom wake words and command processing.

1. Starting a Voice Assistant Design

This video explains how to log into the Voice Assistant Designer, provides an overview of the User Interface and explains wake words, text, variables and intent when building a voice model.

Open resource
Open in media player

2. Creating a basic voice model using Voice Assistant Cloud Tool

Create a new project. Build a Simple Command Structure, Generate a Model from your Project.

Open in media player

3. Testing Your Model in the Cloud

Testing your model in the cloud.

Open in media player

4. Integrating the generated model into your embedded application

Download the generated model. Include the model in your application. Compile and test your embedded application.

Open in media player

5. Testing your embedded model in more detail

Download the generated model. Include the model in your application. Test your embedded application in more detail.

Open in media player

6. Adding phrasing variation and variables

Add phrasing variation and variables to your command model. Test the phrasing variation and variables in the cloud test interface.

Open in media player

DEEPCRAFT™ Studio: An end-to-end platform for ML on edge devices. Streamlines the entire workflow from data collection and annotation to model deployment. Handle all stages efficiently—data management, model building, evaluation, and edge deployment—in one comprehensive solution.

1. Start your Audio Enhancement design

This video explains the basic hardware and software of Audio Enhancement  and the basics for including it in your audio designs.

Open resource
Open in media player

2. Using AFE Configurator to change settings

Loading AFE Configurator settings, modifying, saving settings and viewing the result.

Open in media player

3. Recording and inspecting audio from DEEPCRAFT(tm) Audio Enhancement

Recording audio using AFE Configurator. Playing back and viewing the result in Audacity.

Open in media player

4. Syncing Settings With AFE Configurator

Connect to a running device. Sync settings to the device. Return settings to the defaults by syncing from the device to AFE Configurator.

Open in media player

DEEPCRAFT™ Studio: An end-to-end platform for ML on edge devices. Streamlines the entire workflow from data collection and annotation to model deployment. Handle all stages efficiently—data management, model building, evaluation, and edge deployment—in one comprehensive solution.

1. What is Model Converter

DEEPCRAFT™ Model Converter is a comprehensive solution designed to facilitate the conversion of pre-trained models for deployment on an Infineon target platforms. By leveraging Model Converter, users can generate code for existing models, optimize pre-trained models for specific target devices through configurable optimization parameters, and validate model performance on the desktop prior to actual device deployment.

Open resource

2. How to use it

DEEPCRAFT™ Model Converter is available in both Graphical User Interface (GUI) and Command-Line Interface (CLI) versions, providing flexibility and convenience for developers working in a variety of workflows.

Open in media player

3. How to deploy in ModusToolbox Sample application

This section explains how to deploy the code generated from DEEPCRAFT™ Model Converter onto the PSOC™ Edge boards using the PSOC™ Edge MCU: Machine learning DEEPCRAFT™ profiler code example.

Open resource