Objective Accurate speed perception is crucial for tasks such as man-controlled rendezvous and docking,and teleoperation of space manipulator.Therefore,it is necessary to conduct in-orbit experiments to explore the influence of long-term spaceflight on human speed perception characteristics.Methods The Time-to-Collision(TTC)paradigm was selected to develop experimental software,using a tablet computer for stimulus presentation.Human speed perception characteristics were evaluated based on the subjects'keystroke response data on the keyboard.Through ground-based experiments,the usability and reliability of the paradigm were explored,and the gravity internal model effect was quantitatively analyzed.Through in-orbit experiments on space station tasks,the influence of long-term spaceflight on human speed perception characteristics was further investigated.Results Under the 1G environment on the ground,the TTC paradigm has a high test-retest reliability(r>0.8),and indicators such as average deviation rate and absolute value of average deviation rate show no practice effect.In addition,ground experiments found that compared to vertical upward movement,vertical downward movement is estimated to be faster(i.e.,keystroke time is advanced),showing the existence of the gravity internal model effect.In the microgravity environment of spaceflight,there are no significant differences in average deviation rate and absolute value of average deviation rate among three stages(pre-flight,in-flight,post-flight)and seven tests,indicating that no obvious changes in astronauts'speed perception ability were found at the existing test time points and paradigms.However,the gravity internal model effect(difference between vertical downward and vertical upward)showed a trend of fading in the early stage of astronauts entering orbit.Conclusion Based on the computer screen TTC estimation paradigm,no significant changes in human speed perception ability were found during long-term spaceflight,but microgravity may weaken the human brain's gravity internal model.