Extended Kalman Filtering
Overview
Extended Kalman filtering (EKF) extends recursive Kalman estimation to nonlinear systems by linearizing their dynamics and measurements around the current state estimate. It is widely used in navigation, robotics, tracking, and online monitoring, including IMU–GNSS and LiDAR–inertial fusion. Common approaches include error-state and manifold-aware filters, adaptive covariance scheduling, robust or constrained updates, and learned models that augment or correct the filter; evaluations often use localization, tracking, and dead-reckoning sequences under noise, sensor outages, or changing conditions. A central challenge is balancing computationally efficient updates against linearization error and miscalibrated uncertainty, which can cause drift or divergence and motivates methods that improve robustness without sacrificing consistency.
Latest papers
27 papers3 papers in the last four weeks, level with the four weeks before. 0.0% of all new papers.