Grasping System Based on Adaptive Low-light Image Enhancement

In the tea packaging industry, traditional manual labor for packaging products is not only time-consuming and labor-intensive but also inefficient. Robot grasping technology is revolutionizing the intelligent manufacturing industry by replacing manual labor and simplifying tea packaging production. However, the success and accuracy of intelligent grasping systems largely depend on the quality of visual input and they are susceptible to light interference. To address this challenge, we propose an advanced robot grasping system. The system enhances the captured images under low-light conditions by adopting an adaptive low-light enhancement algorithm, which improves the success rate of object detection. Based on the HSV color space, the system accurately segments the minimum bounding box of tea packaging, and then uses a calibration algorithm to obtain the transformation matrix between robot coordinates and pixel coordinates, which converts the center coordinates to robot coordinates. Finally, the system employs a grasping strategy for object grasping. Extensive experiments have demonstrated the outstanding effectiveness of our proposed method, providing a promising solution for achieving consistent and accurate robot grasping in various low-light environments.