In the lunar activities of the manned lunar landing project, it is necessary to calculate the astronauts’ somatosensory lunar driving operation according to the space mission verification requirements and the astronauts’ extravehicular operation training requirements. Considering the complexity and coupling among astronauts, lunar rovers, and the lunar environment, the measured driving data on the Earth’s surface cannot be applied to astronaut driving tasks on the lunar surface, and the control behaviors caused by different factors have limited adaptability to the lunar surface terrain and special mechanical environment. Under these circumstances, it is difficult to simulate the somatosensory effects of lunar rover driving training in the lunar environment. In response to the above problems, the astronaut–driving somatosensory calculation model is first described through the principle of human motion perception, the lunar environment is created in a virtual environment, the lunar gravity conditions are set, and a dynamic model of the astronaut‐lunar rover‐lunar surface system is created. Then, a dynamic calculation is used to provide parameters for the astronaut somatosensory calculation model, and the driving somatosensory model is simulated considering the astronauts’ own posture adjustment while driving. Finally, the experiment is designed and analyzed to determine the primary and secondary order of influence on the astronauts’ ability to drive somatosensory activity. The results show that the established driving motion perception model can provide an information model and data basis for astronauts’ lunar surface driving motion perception simulation training and can provide support for the development of manned lunar rovers and astronauts’ lunar surface simulation training.