Precise Firing Methods for Small-caliber Unmanned Intelligent Firearms Based on BP Neural Network and Model Predictive Control

Abstract The firing accuracy of small-caliber unmanned intelligent weapons has an important impact on killing effectiveness. In order to improve the firing accuracy of small-caliber unmanned intelligent weapons against moving humanoid targets, hit probability estimation and image servo control are investigated in this paper. Firstly, a method to predict the impact points based on supervised learning is proposed. An corrected ballistic trajectory model is applied to generate training samples, and the mapping relationship that expresses the state of the impact points is formed. Then, a bullet dispersion acquisition experiment is carried out, and a model of bullet dispersion based on a two-dimensional Gaussian distribution is established by analyzing the distribution characteristics of the impact points. Meanwhile, the estimation method of hit probability applied to image detection is formed. In addition, an image servo control method based on model predictive control and human key point detection is proposed, which effectively improves the tracking accuracy of moving humanoid targets. Finally, numerical simulation and firing experiment validation are carried out, and the results show that the precision firing method of small-caliber unmanned intelligent weapons proposed in this paper can effectively improve the firing accuracy of small-caliber unmanned intelligent weapons.