DEEPCRAFT™ Studio Deploy Model: Radar

This code example demonstrates how to deploy a Machine Learning (ML) model generated from DEEPCRAFT™ Studio on PSOC™ 6 MCU family devices.

The code example collects radar data from the XENSIV™ 60 GHz BGT60TR13C sensor, which is then sent to the ML model to detect specific gestures, such as push or circle. It uses the model.c/h files generated from within DEEPCRAFT™ Studio directly. New models based on the Gesture Detection project can be dropped into the project as-is.

Note: This example does not use the ModusToolbox™ Machine Learning Configurator or any device-specific optimizations.

View this README on GitHub.

Provide feedback on this code example.

Requirements

ModusToolbox™ v3.4 or later (tested with v3.4)
Board support package (BSP) minimum required version: 4.0.0
Programming language: C
Associated parts: All PSOC™ 6 MCU parts

Supported toolchains (make variable 'TOOLCHAIN')

GNU Arm® Embedded Compiler v11.3.1 (GCC_ARM) – Default value of TOOLCHAIN

Supported kits (make variable 'TARGET')

PSOC™ 6 AI Evaluation Kit (CY8CKIT-062S2-AI) – Default value of TARGET

Hardware setup

This example uses the board's default configuration. See the kit user guide to ensure that the board is configured correctly.

Software setup

See the ModusToolbox™ tools package installation guide for information about installing and configuring the tools package.

Install a terminal emulator if you do not have one. Instructions in this document use Tera Term.

This example requires no additional software or tools.

Using the code example

Create the project

The ModusToolbox™ tools package provides the Project Creator as both a GUI tool and a command line tool.

Use Project Creator GUI

Open the Project Creator GUI tool

There are several ways to do this, including launching it from the dashboard or from inside the Eclipse IDE. For more details, see the Project Creator user guide (locally available at {ModusToolbox™ install directory}/tools_{version}/project-creator/docs/project-creator.pdf)
On the Choose Board Support Package (BSP) page, select a kit supported by this code example. See Supported kits

Note: To use this code example for a kit not listed here, you may need to update the source files. If the kit does not have the required resources, the application may not work
On the Select Application page:

a. Select the Applications(s) Root Path and the Target IDE

Note: Depending on how you open the Project Creator tool, these fields may be pre-selected for you

b. Select this code example from the list by enabling its check box

Note: You can narrow the list of displayed examples by typing in the filter box

c. (Optional) Change the suggested New Application Name and New BSP Name

d. Click Create to complete the application creation process

Use Project Creator CLI

The 'project-creator-cli' tool can be used to create applications from a CLI terminal or from within batch files or shell scripts. This tool is available in the {ModusToolbox™ install directory}/tools_{version}/project-creator/ directory.

Use a CLI terminal to invoke the 'project-creator-cli' tool. On Windows, use the command-line 'modus-shell' program provided in the ModusToolbox™ installation instead of a standard Windows command-line application. This shell provides access to all ModusToolbox™ tools. You can access it by typing "modus-shell" in the search box in the Windows menu. In Linux and macOS, you can use any terminal application.

The following example clones the "mtb-example-ml-deepcraft-deploy-radar" application with the desired name "DeployRadarModel" configured for the CY8CKIT-062S2-AI BSP into the specified working directory, C:/mtb_projects:

project-creator-cli --board-id CY8CKIT-062S2-AI --app-id mtb-example-ml-deepcraft-deploy-radar --user-app-name DeployRadarModel --target-dir "C:/mtb_projects"

The 'project-creator-cli' tool has the following arguments:

Argument	Description	Required/optional
`--board-id`	Defined in the field of the BSP manifest	Required
`--app-id`	Defined in the field of the CE manifest	Required
`--target-dir`	Specify the directory in which the application is to be created if you prefer not to use the default current working directory	Optional
`--user-app-name`	Specify the name of the application if you prefer to have a name other than the example's default name	Optional

Note: The project-creator-cli tool uses the git clone and make getlibs commands to fetch the repository and import the required libraries. For details, see the "Project creator tools" section of the ModusToolbox™ tools package user guide (locally available at {ModusToolbox™ install directory}/docs_{version}/mtb_user_guide.pdf).

Open the project

After the project has been created, you can open it in your preferred development environment.

Eclipse IDE

If you opened the Project Creator tool from the included Eclipse IDE, the project will open in Eclipse automatically.

For more details, see the Eclipse IDE for ModusToolbox™ user guide (locally available at {ModusToolbox™ install directory}/docs_{version}/mt_ide_user_guide.pdf).

Visual Studio (VS) Code

Launch VS Code manually, and then open the generated {project-name}.code-workspace file located in the project directory.

For more details, see the Visual Studio Code for ModusToolbox™ user guide (locally available at {ModusToolbox™ install directory}/docs_{version}/mt_vscode_user_guide.pdf).

Arm® Keil® µVision®

Double-click the generated {project-name}.cprj file to launch the Keil® µVision® IDE.

For more details, see the Arm® Keil® µVision® for ModusToolbox™ user guide (locally available at {ModusToolbox™ install directory}/docs_{version}/mt_uvision_user_guide.pdf).

IAR Embedded Workbench

Open IAR Embedded Workbench manually, and create a new project. Then select the generated {project-name}.ipcf file located in the project directory.

For more details, see the IAR Embedded Workbench for ModusToolbox™ user guide (locally available at {ModusToolbox™ install directory}/docs_{version}/mt_iar_user_guide.pdf).

Command line

If you prefer to use the CLI, open the appropriate terminal, and navigate to the project directory. On Windows, use the command-line 'modus-shell' program; on Linux and macOS, you can use any terminal application. From there, you can run various make commands.

For more details, see the ModusToolbox™ tools package user guide (locally available at {ModusToolbox™ install directory}/docs_{version}/mtb_user_guide.pdf).

Operation

Connect the board to your PC using the provided USB cable through the KitProg3 USB connector. The example is set up to use the CY8CKIT-062S2-AI kit.
Open a terminal program and select the KitProg3 COM port. Set the serial port parameters to 8N1 and 115200 baud
Program the board using one of the following:
Using Eclipse IDE
1. Select the application project in the Project Explorer
2. In the Quick Panel, scroll down, and click <Application Name> Program (KitProg3_MiniProg4)
In other IDEs

Follow the instructions in your preferred IDE.
Using CLI

From the terminal, execute the make program command to build and program the application using the default toolchain to the default target. The default toolchain is specified in the application's Makefile but you can override this value manually:
```
make program TOOLCHAIN=<toolchain>
```
Example:
```
make program TOOLCHAIN=GCC_ARM
```
After programming, the application starts automatically. Confirm that "DEEPCRAFT Radar Model Example" is displayed on the UART terminal

Figure 1. Terminal output on program startup
Perform push or circle gestures and observe that the model detects and reports the correct activity. For proper detection, orient the sensor on the board in the same way in which it is trained. Figure 2 and Figure 3 show an example video on how to perform the gestures

Figure 2. Performing a push gesture

Figure 3. Performing a circle gesture

Debugging

You can debug the example to step through the code.

In Eclipse IDE

Use the <Application Name> Debug (KitProg3_MiniProg4) configuration in the Quick Panel. For details, see the "Program and debug" section in the Eclipse IDE for ModusToolbox™ user guide.

Note: (Only while debugging) On the CM4 CPU, some code in main() may execute before the debugger halts at the beginning of main(). This means that some code executes twice – once before the debugger stops execution, and again after the debugger resets the program counter to the beginning of main(). See PSOC™ 6 MCU: Code in main() executes before the debugger halts at the first line of main() to learn about this and for the workaround.

In other IDEs

Follow the instructions in your preferred IDE.

Design and implementation

In this example, the firmware reads the data from a radar sensor (XENSIV™ 60 GHz BGT60TR13C) to detect gestures.

The application captures data from a radar sensor at 200 Hz input frequency. Once at least 10 frames of data have been captured, the inference engine runs to determine the gesture that is occurring. Each possible gesture is assigned a confidence score. The confidence score output from the inference engine is then displayed on the UART terminal. Currently, the radar configuration settings used in this application match the radar's 'Default' use case configuration from the DEEPCRAFT™ Studio. To configure the radar for other use cases, see the radar_settings.h file.

Note: The radar model provided in this example is not production ready and is provided here for reference purpose only. You are expected to develop your own model and replace the files here to quickly test the model.

The model.c/h files can be updated to use other models generated by the DEEPCRAFT™ Studio. The current model detects simple push and circling gestures using a radar sensor. The data is collected using the tensor streaming protocol (v2) and a PSOC™ 6 AI Evaluation Kit with the default settings for the Radar (see Collecting and labeling data using Radar Sensor). Machine learning: DEEPCRAFT™ Streaming Protocol can be used to capture additional sensor data for training the model or create a new dataset for a different type of model. Open DEEPCRAFT™ Studio to make changes and experiment further.

Note: The radar models of large size or higher frame rate may not work as expected in this application by simply replacing the model.c/h files. The mechanism of feeding the radar data to the IMAI_enqueue() function model may need to be adjusted in this application to match the model use case.

Resources and settings

Table 1. Application resources

Resource	Alias/object	Purpose
UART (HAL)	cy_retarget_io_uart_obj	UART HAL object used by retarget-io for the debug UART port
SPI (HAL)	spi_obj	SPI HAL object used to communicate with the radar

Related resources

Resources	Links
Application notes	AN228571 – Getting started with PSOC™ 6 MCU on ModusToolbox™ AN215656 – PSOC™ 6 MCU: Dual-CPU system design
Code examples	Using ModusToolbox™ on GitHub
Device documentation	PSOC™ 6 MCU datasheets PSOC™ 6 technical reference manuals
Development kits	Select your kits from the Evaluation board finder
Libraries on GitHub	mtb-pdl-cat1 – Peripheral Driver Library (PDL) mtb-hal-cat1 – Hardware Abstraction Layer (HAL) library retarget-io – Utility library to retarget STDIO messages to a UART port
Middleware on GitHub	psoc6-middleware – Links to all PSOC™ 6 MCU middleware
Tools	ModusToolbox™ – ModusToolbox™ software is a collection of easy-to-use libraries and tools enabling rapid development with Infineon MCUs for applications ranging from wireless and cloud-connected systems, edge AI/ML, embedded sense and control, to wired USB connectivity using PSOC™ Industrial/IoT MCUs, AIROC™ Wi-Fi and Bluetooth® connectivity devices, XMC™ Industrial MCUs, and EZ-USB™/EZ-PD™ wired connectivity controllers. ModusToolbox™ incorporates a comprehensive set of BSPs, HAL, libraries, configuration tools, and provides support for industry-standard IDEs to fast-track your embedded application development

Other resources

Infineon provides a wealth of data at www.infineon.com to help you select the right device, and quickly and effectively integrate it into your design.

Document history

Document title: CE241419 – DEEPCRAFT™ Studio Deploy Model: Radar

Version	Description of change
1.0.0	New code example

All referenced product or service names and trademarks are the property of their respective owners.

The Bluetooth® word mark and logos are registered trademarks owned by Bluetooth SIG, Inc., and any use of such marks by Infineon is under license.

PSOC™, formerly known as PSoC™, is a trademark of Infineon Technologies. Any references to PSoC™ in this document or others shall be deemed to refer to PSOC™.

© Cypress Semiconductor Corporation, 2025. This document is the property of Cypress Semiconductor Corporation, an Infineon Technologies company, and its affiliates ("Cypress"). This document, including any software or firmware included or referenced in this document ("Software"), is owned by Cypress under the intellectual property laws and treaties of the United States and other countries worldwide. Cypress reserves all rights under such laws and treaties and does not, except as specifically stated in this paragraph, grant any license under its patents, copyrights, trademarks, or other intellectual property rights. If the Software is not accompanied by a license agreement and you do not otherwise have a written agreement with Cypress governing the use of the Software, then Cypress hereby grants you a personal, non-exclusive, nontransferable license (without the right to sublicense) (1) under its copyright rights in the Software (a) for Software provided in source code form, to modify and reproduce the Software solely for use with Cypress hardware products, only internally within your organization, and (b) to distribute the Software in binary code form externally to end users (either directly or indirectly through resellers and distributors), solely for use on Cypress hardware product units, and (2) under those claims of Cypress's patents that are infringed by the Software (as provided by Cypress, unmodified) to make, use, distribute, and import the Software solely for use with Cypress hardware products. Any other use, reproduction, modification, translation, or compilation of the Software is prohibited.
TO THE EXTENT PERMITTED BY APPLICABLE LAW, CYPRESS MAKES NO WARRANTY OF ANY KIND, EXPRESS OR IMPLIED, WITH REGARD TO THIS DOCUMENT OR ANY SOFTWARE OR ACCOMPANYING HARDWARE, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE. No computing device can be absolutely secure. Therefore, despite security measures implemented in Cypress hardware or software products, Cypress shall have no liability arising out of any security breach, such as unauthorized access to or use of a Cypress product. CYPRESS DOES NOT REPRESENT, WARRANT, OR GUARANTEE THAT CYPRESS PRODUCTS, OR SYSTEMS CREATED USING CYPRESS PRODUCTS, WILL BE FREE FROM CORRUPTION, ATTACK, VIRUSES, INTERFERENCE, HACKING, DATA LOSS OR THEFT, OR OTHER SECURITY INTRUSION (collectively, "Security Breach"). Cypress disclaims any liability relating to any Security Breach, and you shall and hereby do release Cypress from any claim, damage, or other liability arising from any Security Breach. In addition, the products described in these materials may contain design defects or errors known as errata which may cause the product to deviate from published specifications. To the extent permitted by applicable law, Cypress reserves the right to make changes to this document without further notice. Cypress does not assume any liability arising out of the application or use of any product or circuit described in this document. Any information provided in this document, including any sample design information or programming code, is provided only for reference purposes. It is the responsibility of the user of this document to properly design, program, and test the functionality and safety of any application made of this information and any resulting product. "High-Risk Device" means any device or system whose failure could cause personal injury, death, or property damage. Examples of High-Risk Devices are weapons, nuclear installations, surgical implants, and other medical devices. "Critical Component" means any component of a High-Risk Device whose failure to perform can be reasonably expected to cause, directly or indirectly, the failure of the High-Risk Device, or to affect its safety or effectiveness. Cypress is not liable, in whole or in part, and you shall and hereby do release Cypress from any claim, damage, or other liability arising from any use of a Cypress product as a Critical Component in a High-Risk Device. You shall indemnify and hold Cypress, including its affiliates, and its directors, officers, employees, agents, distributors, and assigns harmless from and against all claims, costs, damages, and expenses, arising out of any claim, including claims for product liability, personal injury or death, or property damage arising from any use of a Cypress product as a Critical Component in a High-Risk Device. Cypress products are not intended or authorized for use as a Critical Component in any High-Risk Device except to the limited extent that (i) Cypress's published data sheet for the product explicitly states Cypress has qualified the product for use in a specific High-Risk Device, or (ii) Cypress has given you advance written authorization to use the product as a Critical Component in the specific High-Risk Device and you have signed a separate indemnification agreement.
Cypress, the Cypress logo, and combinations thereof, ModusToolbox, PSoC, CAPSENSE, EZ-USB, F-RAM, and TRAVEO are trademarks or registered trademarks of Cypress or a subsidiary of Cypress in the United States or in other countries. For a more complete list of Cypress trademarks, visit www.infineon.com. Other names and brands may be claimed as property of their respective owners.

License

Infineon/mtb-example-ml-deepcraft-deploy-radar

About

Resources

License

Stars

Watchers

Forks

Releases

Packages 0

Languages

Packages