• State TI Thinks Resolved
  • Locked Locked
  • Replies 3 replies
  • Answers 1 answer
  • Subscribers 30 subscribers
  • Views 1341 views
  • Users 0 members are here
Support feedback
Options
Options
Related

This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

IWR6843AOPEVM: Enabling TinyML in embedded mmWave radar systems

Tuan Trinh
Prodigy 135 points
Part Number: IWR6843AOPEVM
Other Parts Discussed in Thread: DCA1000EVM, AWR1243BOOST, , IWR6843AOP, IWR6843, AWR1642

Hi TI Customer Support, 

I have a question:

I am working on AWR1243BOOST and DCA1000EVM capture board. I built AI applications on my computer, but AWR1243BOOST hasn't MCU to embed an AI model.

I refer to IWR6843AOPEVM IWR6843AOPEVM Evaluation board | TI.com (IWR6843AOP evaluation module for integrated antenna-on-package (AoP) intelligent mmWave sensor) and video DEMO (Enabling TinyML in embedded mmWave radar systems | TI.com Video), which embedded deep learning model (gesture recognition) on MCU mmWave RADAR.

Is IWR6843AOPEVM the best selection for the embedded AI model?

Regards,
Tuan Trinh The
  • 0
    TI__Mastermind 20921 points

    Hi Tuan,

    What kind of AI models do you intend to run on the device? We have shown the ability to run simple ANN based models on the device (See Gesture Recognition Demo).

    Tuan Trinh said:
    Is IWR6843AOPEVM the best selection for the embedded AI model?

    Yes, the IWR6843AOP would be a good selection for this but if this is for a product then it's also important to make sure to comply with applicable regulations for the region it will be sold. I only mention this since the operating frequency of the device you have been using (AWR1243BOOST) is 76-81Ghz while the operating frequency for the IWR6843 is 60-64Ghz. Just want to make sure you are aware of the difference there. 

    Note, if you wish to debug your applications using Code Composer Studio (CCS) then the MMWAVE-ICBOOST carrier board will be required in addition to the IWR6843AOP EVM.

    Best Regards,

    Josh

  • 0 in reply to Josh Dye
    Prodigy 135 points

    Hi Josh, 

    Thanks for your reply.

    Josh: "What kind of AI models do you intend to run on the device? We have shown the ability to run simple ANN-based models on the device (See Gesture Recognition Demo)."

    I collected raw data from AWR1243BOOST by DCA1000EVM through an Ethernet port and analyzed data (FFT 1-D, Doppler, ...). Therefore, I built an AI model and run real-time on a PC (people-counting, action-recognition, hand gestures classification, ...). My goal is to run on an edge device as: Nvidia jetson nano, or microcontroller..., but AWR1243BOOST hasn't a microcontroller chip. IWR6843AOPEVM or other RADAR (AWR1642 BOOST, ...) has Arm CPU: ARM R4F @ 200MHz chip, embed an AI model on this chip possible?

    Regards,
    Tuan Trinh The

  • 0 in reply to Tuan Trinh
    TI__Mastermind 20921 points

    Hi Tuan,

    As I stated, we have shown the device is capable of running a simple ANN model in our gesture recognition demos. In that demo, custom features are extracted from the radar data  used as inputs to the model to keep the model size smaller. It is hard to say if this will be possible with any arbitrary model as it will heavily depend on the amount of memory the model will occupy and the complexity of the processing required to run the model.

    Best Regards,

    Josh