A gesture recognition based Human-Computer Interaction control system is developed via LabVIEW in this paper. Furthermore, to solve the existing problems of lower precision and poor real-time ability in gesture recognition algorithm, an improved PSO-SVM classification algorithm of hand-gesture recognition is proposed. Firstly, the gesture sample data is collected by using five bending sensor of data glove, then, in order to improve the recognized precision, the data collected is preprocessed and the optimize the SVM kernel parameter value is found by using the improved PSO algorithm. Finally, the recognized hand-gesture is divided into numbered 1–11 control status values, and sent to the sample robot controller by wireless transmission module and achieving the realization of sampling robot motion control. The simulation results illustrate the effectiveness of designed method on