ukf-sensor-fusion

To configure UKF sensor fusion in ROS 2, set the alpha, kappa, and beta parameters in the YAML file for the ukf_filter_node to control sigma-point distribution for nonlinear state estimation.

You should use a UKF for state estimation when your robotic system is highly non-linear and EKF linearization is insufficient, as the UKF applies sigma-point distribution for robust nonlinear propagation.

To switch from EKF to UKF in robot_localization, swap the executable to ukf_node in your launch file, keeping the same input structure for odom, imu, and map frames to enable direct comparison.

Yes, the robot_localization UKF explicitly supports diff-drive and other non-linear platforms requiring robust state estimation across varying sensor inputs like odom and imu data.

The ukf_filter_node configuration requires standard frame mappings including map, odom, and base_link to correctly process sensor fusion inputs from varying sources.