Is there a way to influence which features the algorithms uses to create an area map? I am looking into using the ZEDX for localization in robosoccer, where many parts of the scene (robots, viewers, etc.) will often change while certain features (like goalposts) are constant. Is there a way to selectively filter out these stationary features and pass them to the algorithm for mapping, either via image segmentation or some other method? And if yes, can it be done via the zed ros2 wrapper, or do we need to use a python script to do it?
Hi @Ryuzuki1204,
This feature is currently not available in the ZED SDK. However this is an interesting feature that can help for many use cases, I will log it internally as a feature request so we can study how to integrate it in the ZED SDK.
The ZED SDK positional tracking algorithm already verifies that the features belong to “static” objects before saving them to the map, however this does not filter out objects which are static during the map recording, and have moved at a later time or a different run of the algorithm.