SLAM精度的向量加权平均自适应调节研究

Research on Vector Weighted Average Adaptive Adjustment of SLAM Accuracy

针对粒子滤波算法需要大量粒子以提高精度以及重采样导致的粒子多样性缺失的问题,提出一种自调INFO(向量加权平均算法)优化的粒子重组粒子滤波算法。首先,通过向量不同的加权平均规则,使得最优粒子引导粒子集向期望区域移动,以此提高估计精度;其次,实时计算最优粒子附近的粒子密度,当密度大于阈值时,自适应调整迭代次数,实时监测粒子密度,根据此指标引入次优粒子的作用自适应调整粒子集分布;最后,重采样阶段将筛选后保留的粒子与剩余粒子重新组合成新的粒子,以此增加粒子多样性。通过仿真实验检验改进算法在SLAM中的性能,结果表明该算法较标准算法相比,其位姿与路标估计精度更高且鲁棒性更佳。

Aiming at the problem that particle filter algorithm needs a sea of particles to improve accuracy and the loss of particle diversity caused by resampling,a particle recombination particle filter algorithm optimized by auto-adjustment INFO(weighted mean of vectors algorithm)is proposed.Firstly,the optimal particle guides the particle set to move to the desired region through the different weighted average rules of vectors,so as to improve the estimation accuracy;Secondly,the particle density around the optimal particle is calculated in real time.When the density is higher than the threshold,the number of iterations is adjusted adaptively,and the particle density is monitored in real time.According to this index,the sub-optimal particle is introduced to adjust the particle set distribution adaptively;Finally,the resampling stage recombines the retained particles and the remaining particles into new particles to increase particle diversity.Simulation experiments are tested to verify the performance of the improved algorithm in SLAM.The results show that compared with the norm algorithm,the algorithm has higher accuracy and better stability in the calculation of pose and landmark.