This paper proposed a robot active olfaction positioning algorithm for the situation of the vehicle traveling in the tunnel carrying dangerous goods (i.e., natural gas) occur a leak led to the proliferation of natural gas. Assignment method is used to search the gas randomly, and evolution gradient search algorithm is used to track the gas, these two could complement each other to ensure the mobile robots move along the plume from lower concentration to higher concentration and eventually locate the position of leakage. Plume of gas diffusion physical environment using a mathematical model based on Fluent, the turbulent flow are based on Reynolds timeaveraged equations and standard kε equation, SIMPLEC algorithm is used to solve the NS equation and the component transportation is used to solve the problem of multicomponent fluid diffusion process. Upon completion of the establishment of twodimensional dynamic model of the plume, and then the plume data was put into Matlab. Combine the data generated from plume model with analog mobile robot to search for the odor source and confirm the odor source by using active olfaction positioning algorithm the design proposed. This design provides a reference to minimize the loss and take effective rescue measures in the accident of transportation of dangerous goods suddenly leakage in a tunnel.