What problem does it solve? Fusing wheel odometry and IMU data on robots with nonlinear dynamics (Ackermann steering, aggressive maneuvering) causes EKF linearization errors or divergence, and engineers need a drop-in Unscented Kalman Filter configuration without rewriting their localization stack. ## Core Features & Use Cases - UKF vs EKF Decision Guidance: Explains sigma-point approximation versus Jacobian linearization and when each filter is appropriate for ground robots. - Complete UKF Configuration: Provides a full ukf_filter_node YAML including alpha, kappa, beta sigma-point parameters, frame definitions, odometry/IMU input matrices, and process noise covariance. - Drop-in Launch Replacement: Shows how to swap ekf_node for ukf_node in ROS2 launch files and run both filters side-by-side on recorded bags for comparison. - Use Case: An Ackermann-steering RC car's EKF diverges during rapid turns; switch the executable to ukf_node, keep the identical sensor config, and validate against the EKF trace in PlotJuggler. ## Quick Start Configure a robot_localization UKF node that fuses my wheel odometry and IMU topics for my Ackermann robot.
