In recent years, robot has been applied to an increasing number of areas. Following the issuance of Made in China 2025 and Industry 4.0, the robot applications are leading to a surge over the whole industries. It is inevitable that the industrial robot will replace humans to work on repetitive and tedious tasks both physically and mentally. The life expectancies of robot components are designed to be long enough to maintain the machine operates in the long run. However, what invariably happens is that, in the specific applications, the whole robot system breaks down due to some unexpected damaged components. Unfortunately, the breakdown will force the factory shut down for a period of time. Thus, it is of utmost importance to predict the occurrence of the robot faults and to conduct the corresponding measures to effectively reduce the odds of machine sudden breakdown during productions, which also offers the further assistance to factories in reducing the hidden potential risks. In this paper, the robot fault diagnosis and prediction system is developed based on our developed WIA-FA wireless sensor networks. A new train of thought on robot fault prediction is put forward for the industrial enterprises as well as the robot manufacturers. Experimental results validate that the system is quite suitable for the robot fault diagnosis and prediction.