60 results about "Non-Gaussianity" patented technology

The invention provides a variational Bayesian adaptive iterative Kalman filtering positioning method and system based on LiDAR and IMU in complex weather, aiming at non-Gaussian measurement noise caused by residual noisy points after LiDAR point cloud denoising in complex weather conditions and limitation of a fixed noise covariance matrix in traditional Kalman filtering. Belongs to the technical field of automatic driving. According to the method, LiDAR feature points and IMU original data are directly fused instead of loosely coupled pose fusion, so that the robustness of state estimation is improved; in order to overcome the problem of negative definite or non-positive definite possibly occurring in a classical adaptive filtering method and meet the requirements of adaptive capacity to time-varying noise, high positioning precision and the like, a variational Bayesian method is adopted to approximate noise point distribution. The method is suitable for the automatic driving technology, the positioning precision under the complex weather condition is remarkably improved, and important technical support is provided for achieving the positioning technology under the complex weather condition.

The invention discloses a self-adaptive vehicle navigation filtering method and device, and belongs to the technical field of intelligent driving. The method comprises the following steps: constructing a nonlinear uncertainty tracking system model according to the mutability of an actual vehicle tracking state, the nonlinearity of a sensor measurement equation and the non-Gaussian property of a measurement error; aiming at a nonlinear uncertainty tracking system model, combining strong tracking filtering and a maximum entropy criterion, and constructing a cost function for estimating the optimal state of the unmanned vehicle; determining a fading factor in the cost function based on the orthogonality of the measurement residual sequence; constructing a novel adaptive navigation filtering algorithm by combining an unscented Kalman filtering framework according to the cost function and the fading factor; and carrying out data processing on the unmanned vehicle tracking system according to the constructed adaptive navigation filtering algorithm. According to the method, the problems of mutability, nonlinearity, non-Gaussian property and the like in a nonlinear tracking system can be inhibited at the same time, and the tracking precision and reliability of the unmanned vehicle in a complex environment are improved.