FastSLAM算法的仿生优化改进研究

Study on Improvement of Bionic Optimization of FastSLAM Algorithm

针对机器人导航标准的快速同步定位与地图构建算法(FastSLAM)在重采样过程中存在采样粒子集的贫化以及粒子多样性的缺失导致机器人的定位与建图的精度下降的问题,提出一种基于改进的蝴蝶算法来优化FastSLAM中的粒子滤波部分。改进的算法将机器人的最新时刻的观测和状态信息融入到蝴蝶算法的香味公式中,并在蝴蝶位置更新的过程加入自适应香味半径和自适应蝴蝶飞行调整步长因子,来减少算法的运算时间以及提高预测精度,同时引入偏差修正指数加权算法对粒子的权值进行优化组合,对组合后部分不稳定的粒子进行分布重采样,保证粒子的多样性。通过仿真验证了该算法在估计精度与稳定性方面优于FastSLAM,因此在移动机器人运动模型的定位与建图中具有较高的定位精度与稳定性。

In the process of resampling,the fast simultaneous localization and mapping algorithm of robot navigation standard(FastSLAM)has problems such as the dilution of sampling particle set and the lack of particle diversity,which leads to the decrease of robot localization and map construction accuracy.Therefore,we propose an improved butterfly algorithm to optimize the particle filtering in FastSLAM.The improved algorithm integrates the latest observation and state information of the robot into the fragrance formula of the butterfly algorithm,and adds the adaptive fragrance radius and the adaptive butterfly flight adjustment step factor in the process of updating the butterfly position to reduce the operation time of the algorithm and improve the pre-operation the measurement accuracy.At the same time,the deviation correction index weighting algorithm is introduced to optimize the combination of the weights of particles,and the distribution resampling of some unstable particles after the combination is carried out to ensure the diversity of particles.Simulation results show that the proposed algorithm is superior to FastSLAM in terms of estimation accuracy and stability,so it has higher positioning accuracy and stability in mobile robot motion model.