Pose estimation of several keypoints from IMU and low freq. spatial data

asked 2017-10-11 03:46:08 -0500

Hendrik Wiese gravatar image

updated 2017-10-11 03:48:38 -0500

Hi everybody,

I'm looking for a way to combine sensor data from IMUs (high frequency, high precision, low trueness) with some spatial data that I get from a low frequency, high trueness, high variance source like in my case reconstructed 3D keypoints from disparity images of a stereo camera. That is, on one hand I get the keypoints as jumpy absolute 3D locations with a high variance and low frequency (about 2 to 3Hz). On the other there's a dedicated IMU for (almost) each of the keypoints. Of the keypoints that I have IMUs for I would like to estimate the location based on the two sensor sources with preferably low variance, high precision, high correctness and high frequency.

I suppose there are already a couple of ROS packages that I could use for this purpose, like for instance robot_localization whose purpose however is more for actual robot applications where I also have an odometry source and where the high trueness spatial data comes from a GPS (navsat) as more global altitude, latitude, longitude instead of more local x, y, z. Are there other packages that are more suitable for my rather abstract needs where the type of the object whose position is to be estimated is less relevant, a point in space instead of a real robot?


edit retag flag offensive close merge delete