In order to improve the fault detection effect of intelligent Chemical robots, this paper design fault detection algorithm of intelligent Chemical robots. Based on the Internet of Things and deep learning feature extraction technology, this paper combines WIFI and TCP/IP to optimize the RSSI path loss model, and obtains a path loss model with multi-path effects suitable for Chemical roadways to construct simplified model of robots. It is found that the robots could climb over vertical barriers with a height of 19 cm. It can be seen that the practicability and reliability of communication system are verified by the experiment. The stability conditions and theoretical barrier height of robots are obtained, which is significantly improved.