I am running 2D cartographer with IMU data on the bag below. (Pure scan matching is already tuned to give decent results.) At some point, due to the mounting of the LIDAR, we are almost blind and need to rely on IMU only. However, in that very situation, the walkway the robot is driving on has a drop in the kerb which leads to a y-acceleration.
I am struggling to find the right parameters to give less weight to the acceleration and more to the rotational IMU data (not using global optimization for now). Can you name them? I already played around with the POSE_GRAPH.optimization_problem.-parameters but without any luck.
christqoh
commented
5 years ago
I am running 2D cartographer with IMU data on the bag below. (Pure scan matching is already tuned to give decent results.) At some point, due to the mounting of the LIDAR, we are almost blind and need to rely on IMU only. However, in that very situation, the walkway the robot is driving on has a drop in the kerb which leads to a y-acceleration.
I am struggling to find the right parameters to give less weight to the acceleration and more to the rotational IMU data (not using global optimization for now). Can you name them? I already played around with the
POSE_GRAPH.optimization_problem.-parameters but without any luck.launch-file.