Hello,
In my project, I am required to evaluate accuracy of rviz map generated using GMapping algorithm on a robotino. I need to develop a repeatable method to extract features (not the 'measure' tool in rviz) and determine the distance between them on the map and compare it to measured real-world distances. How do I go about extracting the features from an rviz map?
Originally posted by sam3891 on ROS Answers with karma: 46 on 2015-01-28
Post score: 0
You can interpret the map as an image and do standard image processing on it, for example using OpenCV. This way, you can run different feature detectors/descriptors on the map and use those. An example of doing exactly that is provided in the mapstitch package that allows aligning maps which each other. The code for converting a map to a cv:Mat is here.
Originally posted by Stefan Kohlbrecher with karma: 24361 on 2015-01-28
This answer was ACCEPTED on the original site
Post score: 1
Comment by sam3891 on 2015-01-28:
Thank you. So I take the stored rviz map and run the code (that converts to a cv:Mat), this will give me an image, then I run image processing algorithms using OpenCV, have I understood it right? What is the usage of the mapstitch package?
Comment by Stefan Kohlbrecher on 2015-01-31:
The mapstitch package performs feature detection on two (different) maps. estimates the relative transform between them and generates a stitched/fused map from both.
Comment by sam3891 on 2015-02-04:
Thanks. I will try it out.
