To mitigate the wastage of particles in the FastSLAMl. 0 algorithm, and improve its accuracy, a new method of obtaining the importance function is presented by the FastSLAM2. 0 algorithm. The new method uses the EKF algorithm to estimate the pose of the mobile robot recursively for obtaining the mean value and variance in each time step which constitute a Gaussian distribution function, which is just the importance function. The importance function includes the historical message of the mobile robot's pose, so the FastSLAM2. 0 algorithm can delay the speed of the wastage of particles. The flow of the FastSLAM2. 0 algorithm is provided, and the result compared with the FastSLAMl. 0 algorithm, proves that the accuracy of the FastSLAM2. 0 algorithm is better than the FastSLAMl. 0 algorithm.%为有效缓解FastSLAM1.0算法中的粒子损耗问题,提高其精度,FastSLAM2.0算法提出了一种求取重要性函数的方法.该方法利用扩展卡尔曼滤波算法对移动机器人的位姿状态进行递归估计,得到各个时刻的位姿状态的估计均值和方差,并由此构建服从高斯分布的重要性函数.该重要性函数包含了机器人位姿的历史信息和最新的观测信息,因此可以延缓粒子损耗速度.给出了FastSLAM2.0算法的具体流程,并将其仿真结果与FastSLAM1.0算法进行比较,结果表明了FastSLAM2.0算法的精度优于FastSLAM1.0算法.
展开▼