AM67A: EDGE-AI-STUDIO-MCU

Hi Kanae,

I have redirected your question to the appropriate engineer.

Thank you,

Fabiana

Kanae said:

Q2. EDGE-AI-STUDIO-MCU does not support Cortex-R5F microcontrollers, but is there any plan to do so?

There are no plans at this time to support the AM6x devices on the MCU desktop tool.   Could you share what interests you about AM67 support on the desktop?  

Kanae said:

Q3. Does EDGE-AI-STUDIO-MCU also only support BYOD tuning of models prepared by TI?

That is correct,  both tools are BYOD at this time, although we are considering some form of BYOM for a future release.  Would you mind sharing what kind of use cases you are interested in being supported in the tool.   

Kanae said:

Q1. When AM67A is used, is a means of communication (e.g. OpenAMP) between the Cortex-A side and the Cortex-R5 side provided?

To use the AM67A in Edge AI Studio the device needs to be running a PROCESSOR SDK  Linux image.   I will ask an expert in the Processor SDK to provide more detailed answer to your question.

Hi Martin,

Thank you for your reply.
Regarding Q2. and Q3., I will check our customer's needs and provide feedback here.
As for Q1, I am waiting for Processor SDK expert's answer.

Best Regards,
Kanae

Hi Kanae,

Kanae said:

Q1. When AM67A is used, is a means of communication (e.g. OpenAMP) between the Cortex-A side and the Cortex-R5 side provided?

Yes, please take a look at the AM67A Linux SDK Documentation for more details about this. Let me know if you have any specific questions about this.

IPC for J722S — Processor SDK Linux for AM67A Documentation

Thank you,

Fabiana

Hi Fabiana,

Thank you for your reply.
I will share it with my customer.

Best Regards,
Kanae

Hi Martin,

I received the following comments about customer use cases and AM67 support
in the EDGE-AI-STUDIO-MCU of interest.

First, we were looking for an AI tool that can develop AI for anomaly detection based on sensor data for the Cortex-A MPU.
Secondly, we thought it would be good if we could develop AI on the Cortex-R side, not on the Cortex-A side.
If the AI tool supports MCUs with Cortex-R, we would develop models targeting those MCUs,
and then port the models during software development on the Cortex-R side of the A67.
Therefore, we expected EDGE-AI-STUDIO-MCU to be able to develop AI targeting Cortex-R.
In addition, considering the required functions on the software side, it is necessary to use Linux,
so the use of an MPU is mandatory, and the use of an MCU is not an option.

Please comment if you have any better suggestions for future development studies based on the customer's situation.

Best Regards,
Kanae

Hi Kanae,

No problem. I will let Martin Imrisek address your last message regarding EDGE-AI-STUDIO-MCU.

Thank you,

Fabiana

Hi Fabiana

Thank you for your support.

I would like to know when I have Martin's comments?

Best Regards,
Kanae

Kanae,

I can look into this for you. Thank you for your patience.

Best Regards,

Fabiana

Currently there is no support for AI development on the Cortex-R cores of the MPU devices.   Cortex-R based MCU devices are not yet supported  for model development on Edge AI Studio.

Hi Martin,

Thank you for your support.
I understand that AI development on the Cortex-R cores of MPU devices is not currently supported,
but are there any plans to continue to support Cortex-R based MCU devices for model development in Edge AI Studio?

Best Regards,
Kanae

Hi Fabiana,

Thank you for your reply.
I will share it with my customer.

Best Regards,
Kanae