목표 움직임 예측을 기반으로 한 그리퍼 자동 그립 제어 시스템 설계

초록

To address the challenge of stable grasping of moving targets in dynamic scenes, this paper proposes an automatic gripper grasping control system based on target motion prediction. The system comprises four modules: visual perception, motion prediction, grasp planning, and execution control. It acquires target states through multi-sensor time synchronization and external parameter calibration, performs short-term trajectory prediction within the field of view, and explicitly models the delay in the perception-execution chain. For mobile target grasp planning, Model Predictive Control (MPC) is employed, integrating end-effector reachability, collision avoidance, and grasp stability assessment to generate time-parameterized grasp trajectories online. The execution layer incorporates impedance-based hybrid force/pose control to enhance robustness against target velocity uncertainties. The system is implemented on ROS2 and a real-time controller. Experiments conducted under random acceleration/deceleration conditions (velocity range: 0.1?1.5 m/s) results show: under the predictive system, the average position prediction error is 4.2 mm, with a grasping success rate of 91.7%, and the system maintains stable performance continuously. The study validates the effectiveness and universality of the integrated “prediction-planning-control” framework in dynamic grasping, providing an engineering-ready solution for high-speed, high-precision autonomous grasping.