This idea seems very cool. I would suggest using arUco markers for pose detection. They can be easily pasted on the robot and system of multiple cameras can be used to detect their position and orientation. This will provide enough accuracy. In fact, this system will be more accurate than odometry based tracking. Feedback from such system can be used to correct odometry error as well.
manpreet
Best Answer
2 years ago
I'm looking for technologies to keep track of a robot in a map driving around on a field (SLAM). I want to do it using image processing since it doesnt require hardware modification.
I can use the traditional way of using openCV and the filters to estimate a position by detecting the lines on the field. However people have been using openCV for these problems for years and I was wondering if there was any other (newer) way than openCV?
The augmented reality technology is really interesting and works so well on my phone. I dont think the technology behind is works like openCV where its just checking for differences in colors for example. Can anyone provide me information about the technology behind it?
I'm going to program in C++/ ROS.