In this paper, a bionic perception method of navel orange plucking position based on Fmincon and Proportional Differential (PD) angle control is proposed to solve the problems of wind disturbance and green branches in dynamic unstructured environment. Different from these algorithms that limited to two-dimensional images, this method realizes picking position perception in three-dimensional. Meanwhile, the perception method and the picking robot control algorithm are achieved simultaneously. Firstly, an optimal solution model of the global target rotation angle of the control system based on Fmincon is established to solve the angle optimization problem of robot target approach motion. Secondly, a bionic perception system of plucking position based on PD angle control is constructed to solve specific perception problems. Finally, a joint simulation platform for picking robots based on Solidworks, Adams, and Simulink is given; the validity and accuracy of the algorithm were verified. The experimental results show that the picking accuracy rate is 95%, the angle error of each mechanism and the displacement error are less than 0.5 degrees and 10mm, respectively. The total time from the optimized angle calculation to the system's stability is only about 0.33s. This method is suitable for the rapid perception of plucking position and active angle control of picking robots under dynamic unstructured environment.