A novel trajectory planning method for 4-DOF high-speed parallel robot was studied. The robot had a single action platform structure to reduce the weight of the end platform, so it had good acceleration and deceleration performance. The inverse kinematics model of the mechanism was established. By combining the advantages of the 345 polynomial and the trapezoidal motion law, a 345-corrected trapezoidal motion law was constructed that can take into account both the movement duration and the motion stability. On this basis, the right-angle part of the traditional gate-shaped trajectory was transitioned through the interpolation superposition trajectory. Meanwhile, the shortest total motion duration and the minimum angular acceleration peak value were considered as the goal to optimize the timing of the interpolation and superposition. Finally, the motion frequency experiment was carried out on the prototype. Results showed that the motion frequency of the three-stage fusion path prototype can be increased to 218 times/min and its work efficiency was 55.7% higher than that obtained by using the traditional Adept gate-shaped path. This research was of great significance as it could enhance the efficiency of parallel robots in capturing lightweight and scattered materials in production lines across various industries such as consumer electronics, chemical engineering, food processing and pharmaceuticals. The effectiveness of the parallel robots in grasping delicate materials that were scattered was significantly boosted, resulting in improved production output.