Based on the Condition Assessment of Thin-walled Robot Bearing, developed an assessment method of multi-sensor data fusion. Firstly, acoustic emission（AE） and vibration acceleration signal were collected, and the characteristics of two kinds of signals were extracted separately in time domain and frequency domain. Secondly, the characteristics of AE and vibration signals were used as the characteristic parameters to identify the fault type of thin-wall robot bearing. Finally, self organizing maps（SOM） neural network was used to integrate the characteristics information of AE and vibration signals. The result shows that the method can effectively identify the fault type of thin-walled robot bearing after analysed of the measured data.