DWM1001-DEV sending a lot of NaN when positioning


I’m using DWM1001-DEV boards to position a human in relation to a robot. I have 3 anchors mounted on the mobile robot, in fixed positions about 1 meter in an equilateral triangle, and 1 tag on the human. I also have a passive anchor connected to my computer, for communication. The active anchors are connected at about 20cm from the robot’s top metal plate. The mobile robot is a Husky (by Clearpath robotics), if it helps.

The issue is: about 90% of the data I get from the passive anchor positions the tag at NaN. Do you have any idea why that is the case? Anything would be helpful…

Thank you very much for your help.

Best regards! :slight_smile:

Hi @lucas.he
the main reason here is probably that the tag need to see at least 3 anchors and human body is a quite big signal blocker. So for RTLS applications is recommended to put the anchors to the ceiling.

Possible scenarios are:
a)one (or more) of anchors could not reach the tag
b) the signals are distorted by human that the internal position estimator could not estimate the position.

It would be helpful if you will connect the the TAG shell and observe the measurement via “les” command. You should see there three distances and the position. If not then something is wrong, post here the les output (just few lines).


Hi @lucas.he , I have the same problem: I want to make an husky following an human. The data about the distance of human should be used by a ros application on husky.

  1. first consideration: the anchors so not return tag positions/distance
  2. an anchor configured as passive sees the tag but most of values returned are nan (as you)
  3. my idea is to put two tag on the husky in a fixed position and put the initiator anchor on the human hands. I should be able to measure the distance from each tag and using trigonometric functions evaluate the position.
    did you solve in different way ?
    Thanks in advance