Anti-tank guided missile system design based on an object detection model and a camera

The second generation of anti-tank guided missiles (ATGMs) depend mainly on the operators' skills to aim and hit targets, which makes them less precise. On the other hand, third-generation ATGMs are costly and susceptible to electronic interference. To address these issues, a smart and autonomous control circuit for ATGMs was designed and built using artificial intelligence and computer vision technologies. The control circuit is accurate and cost-effective. The system uses an object detection model to identify the target tank and employs the fundamental matrix and triangulation methods to determine its location. Therefore, the resulting control circuit is intelligent, automated, precise, and reasonably priced. To detect a tank, two consecutive images of the tank were taken using a camera from NVIDIA Jetson TX2 Development Kit. The You Only Look Once (YOLOv5) model was then utilized to identify the tank and draw a bounding box around it. Next, the normalized 8-point algorithm and triangulation method were employed to determine the tank's location. The ATGM was then guided toward the targeted tank using servomotors. The YOLOv5 object detection model was tested using a toy tank, while the targeted tank's location was tested using a pan and tilt camera. The system's performance was satisfactory. The weapon was tested indoors and successfully installed on a real anti-tank missile. By incorporating object detection models and computer vision technologies into the weapon industry, an intelligent and autonomous weapon was developed.