In order to improve the efficiency of nuclear inspection and emergence robot (NIER), a local path optimization method based on the characteristics of NIER task in nuclear station is designed. Kinematic model of NIER is established and described, and the distribution function of NIER along a line and an arc is derived by using Ito Fokker-Planck equation. A correct distribution is come out after fitting the distribution function with the point clouds from a special experiment. Then an efficiency and security mechanism is presented to perform local path re-optimization. The experiment shows the the proposed method can re-optimize the NIER path and improve the efficiency especially in NIE operation.