基于改进贝叶斯估计的水下地形匹配辅助导航方法

Underwater TMAN method based on improved Bayesian estimation

以多波束测深数据为依据,提出了一种基于贝叶斯估计的智能水下机器人(autonomous underwater vehicle,AUV)水下地形匹配辅助导航方法,针对在地形特征不明显区域该方法伪点增多的不足,引入了费希尔判据准则,提出了一种以最大有向特征参数作为二次判别依据的贝叶斯-费希尔估计方法,该方法有效地减少了伪点的个数,增强了地形平坦区域的识别率,对初始定位误差具有良好的鲁棒性。基于电子海图数据和多波束海上实验数据的仿真定位结果,表明了该方法的有效性和可行性。

According to the depth data measured with multi-beam sonar, an underwater terrain matching aided navigation （TMAN） method based on Bayesian estimation is proposed for autonomous underwater vehicle （AUV）. Aiming at the shortcoming that false peak increases intensely in the areas with non distinct terrain characteristics for this method, the Fisher criterion is introduced, and a Bayesian-Fisher estimation method is proposed, which uses maximal directed characteristic parameter （MDCP） as the second-criteria basis. This method effectively reduces the false peaks, enhances the discrimination of terrain fiat area, and has good robustness for initial positioning error. The simu- lation positioning results based on the digital map data from the electronic chart of a certain sea area and the sea test data obtained from multi-beam sonar show the effectiveness and feasibility of the proposed method.