When using robots to carry out grinding and polishing processing of industrial blades, due to factors such as non-zero approach speed and discontinuous dynamic characteristics, the robot grinding and polishing processing has contact force impact and oscillation problems during the contact transition process, which seriously affects the quality of the blade surface processing, contour accuracy and control system stability. The robot grinding and polishing process needs to interact with the environment and complete the contact-non-contact operation with the blade in real time. Therefore, the contact transition state is an important link to realize the robot’s smooth grinding and polishing. This paper proposes an optimization method for impact suppression in the transition state of robot grinding and polishing. Taking the robot abrasive belt grinding and polishing as the research object, the adaptive weighted particle swarm algorithm is used to optimize the input shaper to realize the parameter self-tuning of the input shaping technology, the optimization process does not need system modeling information and has high accuracy. MATLAB is used for simulation analysis, and the self-built robot grinding and polishing system is used to carry out experiments. The results show that the proposed method can realize the smooth transition of the blade grinding and polishing process, effectively shorten the system stabilization time, reduce the maximum overshoot, and accelerate the system response speed, it has strong stability and flexibility.