Abstract: A new method based on the modified TOPSIS is proposed for the target recognition problem of robot non-vision multi-sensors.According to the membership degrees matrix，the recognition problem of multiple targets is transformed to the multi-object decision making problem.After introduced the entropy weight and the relative approach degree to improve TOPSIS，the rule of object recognition is given.The applied example proves that the method is both effective and practicable.

Key words: robot, multi-sensor information fusion, Technique for Order Preference by Similarity to Ideal Solution（TOPSIS）, entropy weight, relative approach degree

摘要： 针对机器人非视觉多传感器目标识别问题，提出了一种基于改进TOPSIS的多传感器信息融合方法。该方法利用指标隶属度矩阵把多目标识别问题转化为多目标决策问题。通过引入熵权和相对接近度改进TOPSIS，给出多传感器目标识别规则。应用实例验证了算法的有效性和实用性。

关键词: 机器人, 多传感器信息融合, 逼近理想点排序方法, 熵权, 相对接近度