• State TI Thinks Resolved
  • Locked Locked
  • Replies 1 reply
  • Answers 1 answer
  • Subscribers 31 subscribers
  • Views 5310 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.

IWR1642BOOST: Python Implementation for radar signal processing algorithms and demos

MikiStitch
Prodigy 130 points
Part Number: IWR1642BOOST
Other Parts Discussed in Thread: IWR1843, IWR1642

Hello Everyone,

 

We are a team of students in University of Illinois at Urbana-Champaign working on radar signal processing on TI platform. We have developed a Python-based package that can perform most functionalities TI mmWave demos provide. To express our gratitude of TI’s support and make our package better, we have open-sourced it under https://github.com/PreSenseRadar/OpenRadar. It is intended for people who wants to quickly evaluate TI radar’s capabilities or fast-prototyping. All you need is the raw ADC data and then you can [edited by Moderator] strictly focus on algorithm development.

 

Below are some features of our package:

  1. Tested on IWR1642 and IWR1843. Support streaming and offline processing of DCA1000 ADC data and offline processing of TSW1400 ADC data (streaming requires mmWave Studio from TI).
  2. Built on Numpy and most matrix operations are vectorized so it’s highly efficient. For the people counting demo we have implemented, it can achieve more than 25 fps for a laptop CPU.
  3. Most TI’s functions are re-implemented and optimized. We even provide two demos, people tracking and visualizer (similar to IWR1843 demo), to showcase how our functions are intended to be used.
  4. Detailed documentations with functions and hosted on https://openradar.readthedocs.io. We also have tutorials about the basic knowledge of FMCW radar from the perspective of programmers.

 

Here is something we want to add/improve in the near future:

  1. More AoA algorithms support.
  2. Better API design and code formulation.
  3. More tracking algorithms support.
  4. Basic machine learning algorithms showcase.

 

We are exciting to share this with everyone to make both ourselves and the radar community better. While we strive to deliver our prototype as a useable Python package, none of us has profound software development experience. As a result, there are definitely inefficiency or bugs in the code. Please share with us the problems you have found and we will fix them as soon as we can. Furthermore, if you have any suggestion for improvement or interesting demos you want to see, leave an issue in our GitHub page. It would also be great to star us if you find our package useful.