Aiming at the real-time motion planning of the six-link picking manipulator, an improved artificial potential field (IAPF) optimization algorithm was proposed based on traditional artificial potential field (APF) algorithm. The elementary transformation sequence method was used to model the kinematics of the manipulator. This model could be used to easily obtain the Jacobian matrix of each point on the manipulator relative to the base coordinate system. In the IAPF algorithm, the Sigmoid function was added on the basis of the original gravitational potential field calculation model of the APF algorithm to make the path planning algorithm converge; the anti S Sigmoid function was introduced into the repulsion potential field to prevent the repulsion explosion. In the IAPF algorithm, the local minimum trap and the target unreachable problem of traditional algorithms are avoided, and the computation time and joint error of robot motion planning are effectively reduced. Finally, the experimental results verify the real-time and accuracy of the proposed optimization algorithm in the obstacle avoidance motion planning of the picking manipulator.