RpLIDAR unable to see black objects

Question

I am working on a ROSBOT for a natural navigation project.

While working with RpLIDAR A2 version, I found that it does not sense thin black surfaces like black pants/trousers while at close range (less than 30 cms).

The screenshots of the RVIZ visualization of the robot is attached below

When the person at more than 80 cm from the LIDAR

When the person at more than 50 cm from the LIDAR

When the person reaches 30 cm from the LIDAR

I tried debugging the code, but no errors in the code.

What are the methods we can use to improve sensing of the robot, can we do any hardware or software tuning to the RpLIDAR sensor to improve its sensing?

What are the likely errors causing this behaviour?

Answer 1

LIDAR sensors rely on sending out a pulse of light it hitting an object and then traveling back to the sensor to detect it.

Black objects are often problematic because they absorb the light and not enough of the light gets back to the sensor. Depending on the technology that the LIDAR uses this can be more or less of a problem at different ranges but usually you will see errors start to occur near the edges of the working range, near or far depending on which limit your hitting.

The most obvious place to see this is in many of the longer range sensors they will spec the maximum range of the sensor for a given level of reflectivity of the object. Because for long range sensing the observable light bouncing back gets so weak as to be undetectable.

Also, black pants versus a "black" board can have very different surface properties which can change the absorbsion. And also remember that most LIDARs also work in infrared wavelengths so the "blackness" of the object is important in a different part of the spectrum than we can see.

And lastly I believe that the RPLIDAR uses a triangulation method which may be having trouble at closer ranges on oblique surfaces such as pant legs. The incident angle also significantly effects the reflectivity.