Combining AMCL, gmapping, odometry?
Hi folks,
how are things like AMCL, gmapping and odometry combined?
We're building a robot that will be capable of localizing itself within its environment by using a fixed triangulation system (called beacons). This system tells the robot its X and Y coordinates within the map. So the robot knows its current position pretty well.
Orientation however is measured by a gyroscope. Its angular velocity value is integrated over time to extrapolate the robots orientation. It most likely will be drifting.
Besides the gyroscope/accelerometer the robot will also have access to rotary encoders which measure the distance that its wheels will drive.
And the third thing the robot will likely get is a laser scanner. However there might be other moving obstacles in the range of the laser.
Now, how could we combine pretty well known absolute localisation with laser scan, drifting orientation and odometry by means of AMCL and gmapping?
Thanks a lot!
Cheers, Hendrik