• State TI Thinks Resolved
  • Locked Locked
  • Replies 4 replies
  • Answers 1 answer
  • Subscribers 30 subscribers
  • Views 1081 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.

IWR1443BOOST: Detecting Wall Profiles

Connor Kite
Prodigy 30 points
Part Number: IWR1443BOOST
Other Parts Discussed in Thread: IWR1443,

We are evaluating the IWR1443 for use in a drone to enhance obstacle detection and mapping, especially around glass obstacles like windows.

I got the IWR1443BOOST to work with ROS and was getting 3D point cloud data in Rviz.  However, while the visualization captured obstacles like people and chairs, I did not see any indication that walls (dry wall) or windows were being detected.  My assumption is that the walls are being detected, but are being filtered out.  For our application, we would ideally see a point cloud representation of flat wall surfaces that could be used to build a map of a room, like what you get with a LIDAR.

My questions are:

   1. Can the IWR1443 be used to generate a point cloud showing walls like I am describing?  

   2.  Is this something that can be done with Out-of-Box Demo firmware or do I need to dig a bit deeper before I can really evaluate this sensor?

Thank you.

  • 0
    TI__Mastermind 30405 points

    Hi Connor,

    Thanks for the interest in TI mmWave Sensors. This topic has been discussed in the following thread.

    https://e2e.ti.com/support/sensor/mmwave_sensors/f/1023/t/633326

    There's also an experiment named Detecting walls of different materials which provides more information about how different type of walls behave wrt detection at various incident angles.

    To answer your questions:

    1. The IWR1443 point cloud output can be used to detect walls but it is important to note that the point cloud of radar would be sparse as compared to a LIDAR. However, mmWave provides other advantages such as robustness to environmental factors such as dust, fog, rain, low light or glaring sunlight and the ability to detect as well as see behind materials such as drywall, glass and plastics as shown in the above experiment.

    2. You can start your evaluation with the OOB demo using either the ROS visualizer or the OOB visualizer. If using the OOB visualizer, I would recommend you to turn peak grouping off to see all the points. The default settings enable peak grouping which groups neighboring points into one.

    Thanks

    -Nitin 

  • 0 in reply to Nitin Sakhuja
    Prodigy 30 points

    Thank you for your reply Nitin.  

    1.  I have actually seen the experiment you mentioned, which shows results in 1D.  Are there any resources showing wall detection in 2D or 3D?  I am curious about how sparse the point cloud would be.

    2.  I actually was running the ROS visualizer with peak grouping off.  I was able to detect a person as a group of points, and even see movement of individual limbs.  However, I was still not getting readings back for walls, even at close to 90 degrees incidence.  Are there additional parameters or thresholds that could be adjusted to better enable wall detection?  

  • 0 in reply to Nitin Sakhuja
    Prodigy 30 points
    Hi Nitin,

    Per my response, could you point me towards any resources showing wall detection in 2D or, even better, in 3D?
  • 0 in reply to Connor Kite
    TI__Mastermind 30405 points
    Hi Connor,

    Besides the Wall detection experiment shared above, I don't think we have a separate resource that shows 3D point cloud of a wall. The best way to visualize the 3D point cloud would be to use the ROS visualizer demo.

    One of the things I would suggest is to move the sensor w.r.t the wall (even tiny motions like vibrating the hand holding the sensor EVM) and see if the point cloud is visible.

    Thanks
    -Nitin