基于海鸥优化改进采样过程的RBPF-SLAM算法

RBPF-SLAM Algorithm on Seagull Optimization to Improve the Sampling Process

为解决PBRF-SLAM中由于粒子退化和粒子耗尽而导致的定位失真和建图一致性差的问题,提出了基于海鸥优化和最小方差重采样的优化方法。在PRBF-SLAM的采样过程中,采样一系列辅助粒子,并利用海鸥优化算法对这些粒子进行寻优,找到估计位姿的最优解,从而避免因陷入局部极值导致的粒子退化。在PRBF-SLAM的重采样过程中,采用最小方差重采样方法替换原先的重采样方法,充分使用辅助粒子,尽可能保证重采样后粒子的多样性。利用Intel Research Lab和ACES Building公开数据集进行SLAM仿真,结果表明优化后的算法相比Gmapping算法总体的平移误差分别降低了36.36%和41.67%,总体的旋转误差分别降低了33.33%和40%。

In order to solve the problems of positioning distortion and poor mapping consistency caused by particle degradation and particle exhaustion in PBRF-SLAM,an optimization method based on seagull optimization and minimum variance resampling was proposed.In the PRBF-SLAM sampling process,a series of auxiliary particles were sampled,and the seagull optimization algorithm was used to optimize these particles to find the optimal solution for estimating the pose,thereby avoiding particle degradation caused by falling into local extreme values.In the PRBF-SLAM resampling process,the minimum variance resampling method was used to replace the original resampling method,and auxiliary particles were fully used to ensure the diversity of particles after resampling as much as possible.Using Intel Research Lab and ACES Building public data sets for SLAM simulation,the results show that the optimized algorithm reduces the overall translation error of the Gmapping algorithm by 36.36%and 41.67%,and the overall rotation error by 33.33%and 40%,respectively.