• State Resolved
  • Locked Locked
  • Replies 5 replies
  • Subscribers 30 subscribers
  • Views 598 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.

IWR1443: determining height above ground

Andrew Saywell
Prodigy 20 points
Part Number: IWR1443
Other Parts Discussed in Thread: IWR6843ISK, MMWAVEICBOOST, IWR6843ISK-ODS, DCA1000EVM

Hi

1) We are currently using Ultrasonic sensors on large mulching machines (Think Cat D8 Dozer with large spinning head with teeth instead of a blade) to determine head height above ground but are finding that those sensors simply return the height to the first object returning the ping with any other longer time returns ignored. That first object ideally is the ground, but it could be a weed or a small tree which then causes the head to lift. The saving grace is that the lift/lower hydraulic system is slow and as such by the time the machine has moved forward the next return could be ground before the head has substantially moved.

I was thinking that mmwave radar from the demo's Ive seen could generate returns from all objects in the radars FoV  but with some software work I could perhaps track true ground level over things like weeds and small shrubs etc.

I have seen the demo water vs ground but the plot that is shown in the demo, given that you are looking for phase changes associated with water ripples etc doesn't seem to me to be what I would be looking for to determine real ground height.

Is this something that Ti mmwave could be used successfully for?

2) The mulching machine is used to mulch tree stumps that are sticking out of the ground after a tree harvester has cut them off near ground level. Could the mmWave differentiate between ground and wood/stump so that we can see the stumps and use the radar to guide us tracking the tree/stump line? The mulching process generates huge quantities of dust and mulch and as such sight of the stumps can be lost. We use long wave IR Cameras to see the stumps before we can visually see them direct, but even that at time looses sight due to the dust/mulch volumes. I Wonder what effect dust will have on the mmwave radar returns

I would like to get into the eval systems and check this for my self but if it raises flags for you experts then best I avoid the time loss.....

Regards

Andy

  • 0
    TI__Mastermind 23725 points

    Hello

    mmWave sensors will provide you robustness against dirt/dry stock on the ground. Basically it will see through these things. 

    Having said that you might want to try the mmwave sensors in the scenario of your interest to see how it respond to the objects and scene variations.  We would recommend starting with ground detection and then start comparing that to over object  detection. You may want to leverage the fact that content with high moisture will give stronger reflection and use that with other sensing information to make a decision.

    Differentiation you are talking about in the #2 may be a bit difficult with single sensor, this is where cascade configuration can be leveraged. However it would required you to develop your own detection technique from sensors raw data provided by board like.  https://www.ti.com/tool/TIDEP-01012

  • 0 in reply to Vaibhav Mahimkar
    TI__Mastermind 23725 points

    Hello

    Sorry about the response getting truncated here.

    Remainder of the response :

    This would be a Raw require inspection of heat map for data (this can be done on OOB demo for single sensor EVM)

    If data shows visual clues then proceed further to algorithm development. This would involve data based research and analysis.

    There doesn't exist a ready solution to give such detection and object output for highlevel integration.

    Please see IWR6843ISK ODS fofr a good angular resolution and wide FOV EVM.

    Thank you,

    Vaibhav

  • 0 in reply to Vaibhav Mahimkar
    TI__Mastermind 23725 points

    Hello

    Can you please confirm if this helps you.

    Thank you,

    Vaibhav

  • 0 in reply to Vaibhav Mahimkar
    Prodigy 20 points

    Vaibhav


    Thank you for your considered response. I have ordered the module, alone, that you mention for first round of evaluation but some comments Ive notice, which aren't any criticism of you or Ti, just feedback that a newbie to this area has noticed....


    I'm unsure if I can evaluate the module alone, that being the IWR6843ISK-ODS the Ti Page for that module has this comment on it under features "Stand-alone mode for form-factor deployment and testing" but the page has no pdf like the earlier modules that covers OOB evaluation. So I think longer term if it gets past initial testing I'll need the carrier board MMWAVEICBOOST and possibly the DCA1000EVM Is that a correct understanding? That all said I note that the webpages seem to be changing quite a bit in this space as I cover below


    When I looked at the SDK page there were, at the time I looked,  links to a 3.Something software release for release 2 boards and a 2.something software release for release 1 boards. the 6843 seemed to be listed as a release 1 board, but then when looking at the MMWAVE VISUALISER it seemed like I needed the 3.something release of the SDK so I was a bit confused as to what I needed to install. However when I tried to repeat that this morning to compose this email the SDK page has changed heaps with a whole lot of additional offerings and a change of terminology around what the 2. and 3. software is for. I presume I need the 3.04.00.03 software.

    Lastly I notice that the module you have suggested has a wide field of view which I think is possibly counter to what I need. For example if the radiated pattern encompasses +/-60 in az and elev then if the sensor is 2m above the ground then I'm actually covering 5.5m^2 of earth surface which is difficult to get in the space Im interested in looking at. If Im looking for changes in that earth profile  and to keep the sensor away from the head itself so that metal structure doesn't interfere with the radar. Our current Ultrasonic sensors have a +/-15degree cone of detection around their boresight and I have one on each side of the head forward of the center of rotation and to the side far enough that the 15degres misses teh metal sides, though I do note that their operation doesn't seem to be too affected if there is some metal directly in the detection cone as long as none of it is situated so that it returns a ping. I sample to the sides of the head because that is where dust and mulch produced from the actions of the head are least seen. Ideally I would like to sample directly in front of the head, but that's where the dust and mulch are worst. If I understand beam steering correctly that could be used to reduce the effective radiation FoV correct?

    Again I very much appreciate your help and guidance.

  • 0 in reply to Andrew Saywell
    TI__Mastermind 23725 points

    Hello

    Please see inline answers.

    Thank you,

    Vaibhav

    Andrew Saywell said:

    Vaibhav


    Thank you for your considered response. I have ordered the module, alone, that you mention for first round of evaluation but some comments Ive notice, which aren't any criticism of you or Ti, just feedback that a newbie to this area has noticed....


    I'm unsure if I can evaluate the module alone, that being the IWR6843ISK-ODS the Ti Page for that module has this comment on it under features "Stand-alone mode for form-factor deployment and testing" but the page has no pdf like the earlier modules that covers OOB evaluation. So I think longer term if it gets past initial testing I'll need the carrier board MMWAVEICBOOST and possibly the DCA1000EVM Is that a correct understanding? That all said I note that the webpages seem to be changing quite a bit in this space as I cover below

    TI: IWR6843ISK-ODS will allow you to use the demo  (uart data) and see how the point cloud from device looks.

          As you go further in data analysis raw data  ( on LVDS ) can be collected on via DCA1000 board.

    This is where Raw data analysis and matlab processing would be needed.

    Please see the EVM users guide for compatibility with other boards.

     


    When I looked at the SDK page there were, at the time I looked,  links to a 3.Something software release for release 2 boards and a 2.something software release for release 1 boards. the 6843 seemed to be listed as a release 1 board, but then when looking at the MMWAVE VISUALISER it seemed like I needed the 3.something release of the SDK so I was a bit confused as to what I needed to install. However when I tried to repeat that this morning to compose this email the SDK page has changed heaps with a whole lot of additional offerings and a change of terminology around what the 2. and 3. software is for. I presume I need the 3.04.00.03 software.

    TI: PLease see the Tirex OOB demo for the EVM you have which will highlight the necessary SW, Visualiser and SDK version

    https://dev.ti.com/tirex/explore/node?node=AHUVfZLNUKWpA43MPJTM7Q__VLyFKFf__LATEST



    Lastly I notice that the module you have suggested has a wide field of view which I think is possibly counter to what I need. For example if the radiated pattern encompasses +/-60 in az and elev then if the sensor is 2m above the ground then I'm actually covering 5.5m^2 of earth surface which is difficult to get in the space Im interested in looking at. If Im looking for changes in that earth profile  and to keep the sensor away from the head itself so that metal structure doesn't interfere with the radar. Our current Ultrasonic sensors have a +/-15degree cone of detection around their boresight and I have one on each side of the head forward of the center of rotation and to the side far enough that the 15degres misses teh metal sides, though I do note that their operation doesn't seem to be too affected if there is some metal directly in the detection cone as long as none of it is situated so that it returns a ping. I sample to the sides of the head because that is where dust and mulch produced from the actions of the head are least seen. Ideally I would like to sample directly in front of the head, but that's where the dust and mulch are worst. If I understand beam steering correctly that could be used to reduce the effective radiation FoV correct?

    TI: Wider FoV will allow you to get coverage of larger ground surface. Postprocessing will give you pointn cloud information of the objects in that scene and you may choose to ignore the objects outside your area of interest.

    Beam steering can allow you to increase the transmit strength in certain directions and that is used to increase the detection range in required areas. In this case omitting some areas should suffice.



    Again I very much appreciate your help and guidance.