辅助粒子滤波算法改进的UFastSLAM算法

Improved UFastSLAM algorithm based on auxiliary particle filter

UFastSLAM算法的重采样过程会带来"粒子耗尽"问题,为提高算法精度和性能,将辅助粒子滤波算法思想引入UFastSLAM算法,提出了解决SLAM问题的方法——辅助UFastSLAM算法.针对UFastSLAM算法的特点,首先,在其建议分布函数的求取过程中引入最近的观测信息,增强采样粒子权值的稳定性;其次,在其重采样过程中引入1次加权和2次加权,增大观测似然度大的粒子比例以缓解粒子退化问题.实验结果证明,辅助UFastSLAM算法在估计精度、一致性等方面都具有很好的性能.

The UFastSLAM algorithm＇s resampling process will bring ＂sample reproverishment＂ problem. To improve the algorithm performance and increase the estimate accuracy, and based on the thought of auxiliary particle filtering algorithm, a solution named ＂Auxiliary UFastSLAM＂ is presented for SLAM problem. In view of the specialty of UFastSLAM algorithm, the recent observation information in its suggestion distribution function＇s seeking process to enhance stability of particle sampling, and one time weighting and second time weighting in its resampling process to alleviate particle degeneration are introduced. The experimental results indi- cate that the Auxiliary UFastSLAM algorithm does well on both estimate accuracy and consistency.