Abstract: Aiming at the problems of strong randomness, low search efficiency and long path planning time of the bidirectional rapidly-exploring random trees-connect (RRT-Connect) algorithm, an improved RRT-Connect algorithm is proposed. The algorithm sets the third node on the perpendicular bisector of the line connecting the starting point and the target point. By using Gaussian distribution to limit the sampling area of the third node, the path redundancy caused by the third sampling node being far from the midpoint is avoided. The algorithm generates two random trees to the starting point and the target point respectively through the third node, and combines the greedy algorithm idea and introduces the dynamic step method to improve the planning efficiency of the algorithm. The simulation results show that the improved RRT-Connect algorithm reduces the average running time by 48.7%, the average number of iterations by 38.9%, and the average path length by 25.2% compared with the traditional RRT-Connect algorithm. In addition, in view of the accuracy problem of the traditional nine-point calibration method, an improved nine-point calibration method is proposed. This method calculates the length of the sixth axis of the robot arm by obtaining multiple sets of postures of the robot arm at the same point in space. When the joint angles and axis lengths of the robot arm are known, the length of the sixth axis after the end effector of the robot arm is installed is calculated to improve the accuracy of hand-eye calibration. The experimental results show that the improved method has an average accuracy improvement of 2.09% compared with the traditional nine-point calibration method. Finally, the improved RRT-Connect algorithm and the improved nine-point calibration method were verified on the robotic arm platform. The experimental results show that the improved RRT-Connect algorithm reduces the total path planning time and total length by 8.28% and 4.79% respectively compared with the DRRT-Connect algorithm, and the improved nine-point calibration method improves the grasping accuracy by 3% compared with the traditional nine-point calibration method.