论文部分内容阅读
自主式水下机器人(AUV)同时定位与环境地图构建(SLAM)是实现水下未知环境自主导航的关键性技术,是机器人研究的热门课题之一。针对自主式水下机器人SLAM框架下应用扩展卡尔曼滤波(EKF)因模型线性化产生误差以及噪声统计未知的情形,采用一种基于虚拟噪声补偿技术的EKF算法,该方法可以把未知模型误差归入到虚拟噪声中去,运用噪声统计估值器在线估计噪声统计。以构建的AUV运动系统的模型为基准,从滤波精度、收敛性及算法稳定性方面,通过matlab仿真验证改进的EKF算法的效果。仿真结果表明,相对于传统的EKF算法,改进后的EKF算法估计精度更高,预期效果更好,有效提高了非线性滤波的性能。
Autonomous Underwater Vehicle (AUV) Simultaneous Localization and Environment Map Construction (SLAM) is a key technology for autonomous navigation in underwater unknown environment and is one of the hot topics in robotics research. In view of the error of the model linearization and unknown statistics of noise under the application of Extended Kalman Filter (EKF) in the autonomous underwater robot SLAM framework, an EKF algorithm based on the virtual noise compensation technique is proposed, which can return the unknown model error to Into the virtual noise to go, the use of noise statistics estimator to estimate the noise statistics online. Based on the constructed model of AUV motion system, the effectiveness of the improved EKF algorithm is verified by matlab simulation in terms of filtering accuracy, convergence and algorithm stability. The simulation results show that compared with the traditional EKF algorithm, the improved EKF algorithm has higher estimation accuracy and better expected performance, which effectively improves the performance of nonlinear filter.