Filter-Based Secure Dynamic Pose Estimation for Autonomous Vehicles

Pose estimation is a critical problem in autonomous vehicle navigation, especially in circumstances where sensor failure or attacks exist. In this paper, a filter-based secure dynamic pose estimation approach is proposed such that the vehicle pose can be resilient under the possible sensor attacks. Our proposed estimator coincides with the conventional Kalman filter when all sensors on autonomous vehicles are benign. If less than half of the measurement states are compromised by the randomly occurring deception attacks, it still gives stable estimates of the pose states, i.e., an upper bound for the estimation error covariance is guaranteed. The pose estimation results with single and multiple attacks on the testing route validate the effectiveness and robustness of the proposed approach.