Abstract: A coal mine dangerous gas inspection system based on multi-sensor fusion technology is designed. The overall architecture of the system is divided into execution layer, network layer and decision layer. The monitoring platform and server in the decision layer and the mobile inspection robot in the execution layer realize data transmission through the network layer. This paper introduces the process of dangerous gas inspection and the module composition and function classification of inspection robot. The intelligent sensing module of the robot has the corresponding environmental parameter monitoring function. The robot can realize autonomous navigation and obstacle avoidance and dangerous gas concentration data acquisition by carrying a variety of sensors. The BP neural network algorithm is used to realize multi-sensor fusion, The real-time monitoring of various dangerous gas concentrations in coal mine underground environment is completed.