• State TI Thinks Resolved
  • Replies 9 replies
  • Answers 4 answers
  • Subscribers 31 subscribers
  • Views 98 views
  • Users 0 members are here
Support feedback
Options
Options
Similar topics

IWR6843ISK: Angle resolution and object detection

Manoj Wakdikar
Intellectual 425 points
Part Number: IWR6843ISK
Other Parts Discussed in Thread: IWR6843

Tool/software:

Hi,

I am using IWR6843ISK. What is the range, azimuth and elevation angle resolution to differentiate between two different objects? I have found that for range it is 6cm, azimuth 17deg and for elevation 58deg, is it correct please let me know?

If it is, then two different objects are considered as two objects if they are 6cm apart from range point of view, if they are beyond 17deg angle from azimuth angle resolution point of view and similarly beyond 58deg for elevation angle. Let me know if my understanding is correct.

Thanks,

Manoj

  • 0
    TI__Mastermind 33650 points

    See this page for the statistics.

    https://dev.ti.com/tirex/explore/node?a=1AslXXD__2.30.00.12&node=A__AAICr8asHfv7xCnGzOYQhA__radar_toolbox__1AslXXD__2.30.00.12

    Manoj Wakdikar said:

    If it is, then two different objects are considered as two objects if they are 6cm apart from range point of view, if they are beyond 17deg angle from azimuth angle resolution point of view and similarly beyond 58deg for elevation angle. Let me know if my understanding is correct.

    That is correct.

    Best,

    Nate

  • 0 in reply to Nathan Block
    Intellectual 425 points

    Hi Nathan,

    Thank you for your message.

    So if any one of these parameters or properties satisfy their criteria then radar will detect it as two different objects?

    For eg., radar is detecting

    First object:

    Range = 23.98meters, azimuth angle = -6.95, elevation angle = -39.19 and velocity = -0.50

    Second object:

    Range = 12.73meters, azimuth angle = -7.26, elevation angle = 8.69 and velocity = 0.36.

    In above example though angular resolution is very close to each other and does not satisfy it's criteria for azimuth more than 17 deg as well as for elevation angle more than 58 deg but the range is more than 6cm apart, so here the range property satisfies the criteria and hence it is detected as two different objects?

    Thanks,

    Manoj

  • 0 in reply to Manoj Wakdikar
    TI__Mastermind 33650 points

    Yes, the range criteria will separate them.

  • 0 in reply to Nathan Block
    Intellectual 425 points

    Thank you for your inputs. 

    If I point a radar to 20x20 feet big wall. This is the only obstacle in front of radar. What we observed is we are getting few points which are detecting the wall. Since wall is too big so radar should detect multiple points for complete wall, but it didn't. Also, will it be resolved in multiple objects since the range resolution will differentiate that wall after every 6cm distance? Could you please help to understand these scenarios.

    Thanks,
    Manoj

  • 0 in reply to Manoj Wakdikar
    TI__Mastermind 33650 points

    Hi,

    When you point the radar at the wall, it will typically make a couple of points at boresight because that's where the reflections are strongest. The radar doesn't have enough angular resolution to image the face of the wall. If you are running the out-of-box demo, the software will detect a peak in the range-doppler heatmap, and estimate the angle for that peak. If multiple objects are at the same range and doppler (which is the case for a wall), the radar will not be able to distinguish all of them.

    Best,

    Nate

  • 0 in reply to Nathan Block
    Intellectual 425 points

    Hi Nathan,

    I am using "long range people detection".

    If multiple objects are at the same range and doppler (which is the case for a wall), the radar will not be able to distinguish all of them. - if the wall is that big so radar will not be able to differentiate it at every 6cm range? Since, due to angle range will change which can be possibly greater than 6cm of range resolution at some point. Is my understanding correct?

    thanks,
    Manoj

  • 0 in reply to Manoj Wakdikar
    TI__Mastermind 33650 points

    You are correct that the wall will exist in multiple range bins, but they will likely be grouped together by the CFAR operation. If you want to image the wall, you can try with the range-azimuth or range-doppler heat map, but we do not support this application due to the limited angular resolution of the IWR6843 device.

    Best,

    Nate

  • 0 in reply to Nathan Block
    Intellectual 425 points

    Hi Nathan,

    Thank you for clarification. 

    Could you please explain me in detail the limitation of not supporting this application due to the limited angular resolution of the IWR6843 device? I just want to understand it in detail.

    Thanks,
    Manoj

  • 0 in reply to Manoj Wakdikar
    TI__Mastermind 33650 points

    Hi Manoj,

    Since the IWR6843 has 8 antennas in the azimuth dimension, it can only resolve objects that are >14 degrees apart in the azimuth. Since the wall is a large smooth surface, it has elements that are more than 14 degrees apart in the same range bin. The radar will be able to detect the wall, and if you examine it in the radar cube, you will surely see a return at that range and angle, but you will not be able to know where the wall truly starts and ends with a high degree of accuracy, since all the returns from each section of the wall will blur in the azimuth domain. So, you will know there is a barrier there, but you won't know where it necessarily starts and ends in the azimuth domain to a high degree of accuracy.

    Best,

    Nate