Learning vehicle trajectory uncertainty

A novel approach for vehicle tracking using a hybrid adaptive Kalman filter is proposed. The utilizes recurrent neural networks to learn the vehicle’s geometrical and kinematic features, which are then used in supervised learning model determine actual process noise covariance framework. This addresses limitations of traditional linear filters, can suffer from degraded performance due uncertainty trajectory modeling. Our method evaluated compared other filters Oxford RobotCar dataset, has shown be effective accurately determining real-time scenarios. Overall, this implemented estimation problems improve performance.