Abstract: A stabilization control algorithm was proposed for adaptive prediction horizon based on input reconstructing resilient self-triggered model predictive control (ST-MPC) to balance the contradiction between cyber security and resource limitation of robot system. Firstly, based on the self-triggered aperiodic sampling characteristics and a false data injection (FDI) attack model, an input reconstruction mechanism was designed to ensure the robot system can quickly reconstruct and possess a feasible control sequence for weakening the influence of FDI attack. Secondly, combined with the input reconstruction mechanism, a selection condition of key data and prediction horizon regulatory mechanism were designed to reduce resource consumption from two aspects of maximizing the triggering interval and reducing the complexity of optimization problems. Then, a resilient ST-MPC stabilization control algorithm was designed based on input reconstruction and predictive horizon regulatory mechanism, and the feasibility and closed-loop stability conditions of the proposed algorithm under FDI attack were demonstrated. Finally, the proposed algorithm was verified based on simulation experiments to maintain outstanding control performance and resource utilization under the premise of resisting FDI attack.