论文部分内容阅读
提出了一种改进的同时定位与地图创建(SLAM)方法——遗传优化Marginal-SLAM算法用于机器人导航,将Marginal-SLAM算法与遗传算法相结合,继承了Marginal-SLAM算法权值方差较低的优点对粒子集进行优化,进一步提高了算法的综合性能.将地图视为模型的参数,并由递归极大似然估计法得到,位姿估计采用边缘粒子滤波方法求取.设计了一种与Marginal-SLAM算法兼容的遗传算法,融合最新的观测信息来优化粒子集,同时兼顾粒子集的多样性,提高了粒子集的性能.仿真实验表明,该遗传优化Marginal-SLAM算法在有效粒子数和权值方差方面都很好的表现,路径和地图估计的精度也有一定提高.
An improved Marginal-SLAM algorithm is proposed for the simultaneous localization and map creation (SLAM). The Marginal-SLAM algorithm is used for robot navigation. The Marginal-SLAM algorithm is combined with the genetic algorithm. The particle set is optimized to further improve the overall performance of the algorithm.Map is taken as a model parameter and is obtained by the recursive maximum likelihood estimation method and the pose estimation is obtained by using the edge particle filter method. Marginal-SLAM algorithm is compatible with the genetic algorithm, the combination of the latest observation information to optimize the particle set, taking into account the diversity of particle sets, improve the performance of the particle set.Simulation experiments show that the genetic optimization Marginal-SLAM algorithm in the effective particle number And weight variance are good performance, the accuracy of the path and map estimates also have to improve.