search

CCNYRoboticsLab / scan_tools

ROS Laser scan tools
300 stars 187 forks source link

correct sensor fusion calculation #55

Closed seiga-k closed 2 years ago

seiga-k commented 7 years ago

It seems the order of calculation for sensor fusion is wrong. This causes a position and a pose jumping when enables a sensor fusion switch. I only checked with odometry fusion and no fusion mode. But, I have no confidence for a change of line 470. Please confirm well.

EwingKang commented 6 years ago

@seiga-k I've been doubting for the source code for a while. Seems like you have the same problem as I do. Unfortunately, at the first glance, your equation still doesn't seems right to me. Despite the lack of description and fuzzy commenting on the source code, I'm working on derive the equation myself currently. Wish me luck.

EwingKang commented 6 years ago

My main problem is: at laser_scan_matcher.cpp:467 //the predicted change of the laser's position, in the laser frame By "laser frame", does it mean in the frame of last_laser_KeyFrame? Or does it imply the frame that is actually attach to the laser scanner (that is, [f2b + base_tolaser] ) I'll guess it's the latter since line 470 seems like coordinate transformation (but done in a wrong way). However, if that is true, then the line 465 don't make too much of a sense. I'll keep working on it.

EwingKang commented 6 years ago

I've created a new pull request #58 to fix the issue. You can find further information there.

130s commented 2 years ago

Closing as https://github.com/CCNYRoboticsLab/scan_tools/pull/79 seems to be meant to address the same issue. If it doesn't, please feel free to add more info.