结合免疫机制的并发定位与建图多目标进化算法

由于基于进化算法的移动机器人并发定位与建图问题本质上具有多目标特性,因此将该问题转化成为多目标优化问题进行处理.为了有效地提高基于多目标进化算法的移动机器人并发定位与建图方法的效率,提出了结合免疫机制的局部搜索方法.为此,构造了称之为关键点栅格吸引操作的局部搜索方法,它运用了所针对问题的领域知识.通过移动机器人物理实验证实了所构造算法的计算代价比其他基于进化算法的单目标优化方法少,获取的地图准确性更高.

Multi - Objective Evolutionary Algorithms for SLAM with Immunity

The simultaneous localization and mapping problem with evolutionary algorithms is translated to a multi-objective optimization problem since it inherently possesses of multi-objective characters,and in order to efficiently solve the simultaneous localization and mapping problem with multi-objective evolutionary algorithms,a local searcher with immunity is constructed.The local searcher employs domain knowledge of the problem,which is named as a key point grid pulling that is developed in the paper.The experiment results of a real mobile robot indicate that the computational expensiveness of designed algorithms is less than other evolutionary algorithms of single-objection for simultaneous localization and mapping and accuracy of obtained maps are higher.