Open Nigel-BMV opened 2 weeks ago
rostest
commented
2 weeks ago Thank you for your feedback. We will be analysing the behaviour in more detail. Are you using the API on Windows or Linux? Are you using the default settings in the launchfile with full angle range and last echo only? If optional filters are enabled, please post the configuration and the launchfile.
Nigel-BMV
commented
2 weeks ago Are you using the API on Windows or Linux
Windows without ROS
Are you using the default settings in the launchfile with full angle range and last echo only?
The launch file is still in its default state, but I've included it here :
`<?xml version="1.0"?>
`
If optional filters are enabled
There are no optional filters in use, but we are using a distance limiter
Echo Filter = last Echo Distance filter 150-2000mm
rostest
commented
2 weeks ago Thanks for your reply. Unfortunately, I cannot reproduce this error.
Invalid scan points are dropped by default by the rangeFilter settings in the launch file. Depending on the scene, the point cloud dimension (number of valid scan points) may vary when scan points are close to the filter limits. To get constant point cloud dimensions for each measurement, you can disable the rangeFilter in the launch file by setting rangeFilter=0.05,999,0 (rangeFilter off) instead of the default setting rangeFilter=0.05,999,1 (rangeFilter on) in the point cloud configuration for "cloud_unstructured_segments" and "cloud_unstructured_fullframe":
<!-- cloud_unstructured_segments: cartesian coordinates, segmented, all echos, all layers, range filter on, max. 2700 points, mean ca. 1000 points per cloud -->
<param name="cloud_unstructured_segments" type="string" value="coordinateNotation=0 updateMethod=1 echos=0,1,2 layers=1 reflectors=0,1 infringed=0,1 rangeFilter=0.05,999,0 topic=/cloud_unstructured_segments frameid=world publish=1"/>
<!-- cloud_unstructured_fullframe: cartesian coordinates, fullframe, all echos, all layers, range filter on, max. 32400 points, mean ca. 10000 points per cloud -->
<param name="cloud_unstructured_fullframe" type="string" value="coordinateNotation=0 updateMethod=0 echos=0,1,2 layers=1 reflectors=0,1 infringed=0,1 rangeFilter=0.05,999,0 topic=/cloud_unstructured_fullframe frameid=world publish=1"/>Please try the following steps:
rangeFilter=0.05,999,0 (rangeFilter off) in the launch file for the "cloud_unstructured_segments" and "cloud_unstructured_fullframe" settings. The point cloud dimensions should then be constant.Please post the pcapng file for our further analysis.
We are using the picoscan-150 to check a moving object, however we have noticed that occassionally the message data with the -1 ID has sections that are doubled. This messes with the data of the moving object creating a zone with unsure measurements. To get the full scan data, I'm using a filter on the SickScanPointCloudMsg segment_idx, where -1 gives me the full scene.
I have recreated the issue with a static scene. 103 Scans were taken, of which 8 scans have doubled data.
The data has a length of 333 in normal circumstances and visualised appears like :
The doubled data has a length of 393 and visually appears like :
The area which contains the doubled data :
Debug console messaging, as you can see the message is 333 x 1, and 393 x 1 for the following with the first 60 x 1 doubled :
Normal scan field data (X/Y/Z/Intensity)
Doubled Sector scan field data (X/Y/Z/Intensity)
Due to the real setup having objects in motion, meaning a variable amount of points; I am unable to filter this specific situation.
How do I prevent this from occurring, receiving just a singular full scan at all times?