基于模糊逻辑的SLAM数据关联方法

Data association approach for SLAM based on fuzzy logic

提出了一种基于模糊逻辑的同步定位与地图创建(simultaneous localization and maping,SLAM)数据关联方法,该方法计算特征观测和特征估计的误差椭圆,对归一化新息和误差椭圆重叠比例进行模糊化处理,作为模糊系统输入变量,将数据关联结果作为模糊输出变量。通过融合这些特征信息建立模糊规则,进行模糊推理,最终获得数据关联结果。这种方法可以有效表达数据关联中存在的不确定性和模糊性,具有处理多个候选关联假设的能力,并且在实际观测与特征估计值距离较小时减少了误关联的发生,在实际观测与特征估计值距离较大时又避免了丢弃正确关联。仿真实验表明,新算法具有更好的抗干扰能力和鲁棒性,为SLAM的数据关联提供了一条新的途径。

A novel simultaneous localization and mapping （SLAM） data association approach based on fuzzy logic is proposed. It firstly calculates the error ellipses of the feature observation and estimation, and fuzzes up the normalized residue and percentage of the two ellipses＇ overlap. And then these informations are used as the input of fuzzy system variables, in the meanwhile the data association result is defined as a fuzzy output varia- ble. The new algorithm carries out the fuzzy reasoning by fusioning the feature information through the fuzzy rules, and the association result is obtained in the end. This approach describes the uncertainty and fuzziness in the data association validly, and it has the ability to deal with multiple association hypothesis. Further more, it can reduce the false association when the distance between true observation and estimation is close, and it can avoid discarding the true association when the distance is far. The simulation results demonstrate that the proposed approach is of better anti-interference ability and robustness, and it provides a new way for data association of the SLAM problem.