论文部分内容阅读
在人机共存的生产环境中,提高人机协作的安全性对增强工作效率和整体生产力至关重要,为此提出一种基于深度图像检测的实时碰撞避免控制系统。首先,使用两个Kinect图像传感器作为深度摄像机,构建机器人所在环境的3D模型;然后,进行深度图像处理,移除已知固定物体和环境背景;接着,检测并构建操作人员图像的点云,与机器人3D模型合并,计算两者之间最小距离;最后,控制器根据距离阈值做出碰撞避免策略。通过人机装配实验表明,该方案能有效进行碰撞检测和碰撞避免。
In the man-machine coexistence production environment, to improve the safety of human-computer collaboration is crucial to enhancing work efficiency and overall productivity. Therefore, a real-time collision avoidance control system based on depth image detection is proposed. First, two Kinect image sensors are used as depth cameras to build a 3D model of the environment in which the robot is located. Then, depth image processing is performed to remove the known fixed objects and the environment background. Then, the point cloud of the operator image is detected and constructed The robot 3D models are combined to calculate the minimum distance between the two; finally, the controller makes a collision avoidance strategy based on the distance threshold. Man-machine assembly experiments show that the program can effectively detect collision and collision avoidance.