¿EKF works for lidar sensor?

asked 2020-04-23 17:25:49 -0500

Nicolas Cortes gravatar image

Hello everyone

I am working with a differential mobile robot and lidar sensor. I am reviewing the robot_pose_ekf package and I was wondering if it is possible to implement that method with the lidar sensor?

edit retag flag offensive close merge delete


That question is a little nonsensical - please clarify what you mean.

Also, robot_localization has taken over the role of robot_pose_ekf, it has been removed from the navigation stack from Melodic and newer in ROS1 and not ported to ROS2.

stevemacenski gravatar image stevemacenski  ( 2020-04-23 19:28:06 -0500 )edit

Hi Steve, I am doing a project with a turtlebot3 and a 2D lidar sensor. I need to improve the odometry of the robot and I found this method but it seems that it is only for 3D sensors.

Nicolas Cortes gravatar image Nicolas Cortes  ( 2020-04-23 19:31:34 -0500 )edit

These EKFs are used to fuse odometry information into a smoothed state estimation. Asking about how to implement fused odometry from a lidar isn't really a sensible question. You'd need to take that data and translate it first into something that has a type of odometry output to then ask the question about how to fuse multiple odometry sources together. Ex. if I were to ask about how to implement and EKF with a pointcloud, that's a similarly nonsense question because you're comparing apples and oranges.

stevemacenski gravatar image stevemacenski  ( 2020-04-23 20:50:22 -0500 )edit

Could using something like Visual Odometry -- which would take cloud or lidar data as input and outputs odometry estimates -- be something which could be used perhaps? Then feed that information to robot_localization?

gvdhoorn gravatar image gvdhoorn  ( 2020-04-24 04:09:52 -0500 )edit

Hi gvdhoom, I cannot use visual odometry, the robot only has 2D lidar and I cannot add another external sensor. What if the value of vo in EFK is false?

Nicolas Cortes gravatar image Nicolas Cortes  ( 2020-04-24 11:13:37 -0500 )edit