移动机器人的蒙特卡罗自主定位算法研究

Study on Monte Carlo Self-Localization Algorithm for Mobile Robots

在分析基于贝叶斯估计的定位过程基础上,针对Markov定位算法需要大量存贮空间和计算量导致无法在较高频率下有效利用传感器信息的问题,利用基于采样的Monte Carlo方法解决移动机器人的自主定位问题。理论分析和仿真试验表明,Monte Carlo自主定位算法很大程度上提高了定位效率,能够更有效的利用传感信息且降低了传感信息不确定性的影响,在"绑架"后以及在中心对称环境中都表现出良好的全局定位性能。

Global position estimation is a key technique in research of mobile robotics. According to the problem that Markov self-Localization algorithm call for plenty of computing and memory resource, and on the basis of explaining Bayes-ian localization process, we introduce Monte Carlo algorithm that bases on the kernel ideal of sampling. Theoretical analysis and simulation experiments prove that Monte Carlo algorithm improve the positioning efficiency largely, make use of sense information and decrease its uncertainty affect more effectively. After 'Kidnapped' or in symmetrical environment, Monte Carlo self-localization algorithm shows its robust global localization capability.