The use of laser beam welding with robotic manipulators is expanding towards wider industrial applications as the system availability increases with reduced capital costs. Conventionally, laser welding requires high positioning and coupling accuracy. Due to the variability in the part geometry and positioning, as well as the thermal deformation that may occur during the process, joint position and fit-up are not always acceptable nor predictable if simple fixtures are used. This makes the passage from virtual CAD/CAM environment to real production site not trivial, limiting applications where short part preparations are a need like small-batch productions. Solutions that render the laser welding operations feasible for production series with non-stringent tolerances are required to serve a wider range of industrial applications. Such solutions should be able to track the seam as well as tolerating variable gaps formed between the parts to be joined. In this work, an online correction for robot trajectory based on a greyscale coaxial vision system with external illumination and an adaptive wobbling strategy are proposed as means to increase the overall flexibility of a manufacturing plant. The underlying vision algorithm and control architectures are presented; the robustness of the system to poor illumination conditions and variable reflection conditions is also discussed. The developed solution employed two control loops: the first is able to change the robot pose to follow varying trajectories; the second, able to vary the amplitude of circular wobbling as a function of the gap formed in butt-joint welds. Demonstrator cases on butt-joint welds with AISI 301 stainless steel with increased complexity were used to test the efficacy of the solution. The system was successfully tested on 2 mm thick, planar stainless-steel sheets at a maximum welding speed of 25 mm/s and yielded a maximum positioning and yaw-orientation errors of respectively 0.325 mm and 4.5°. Continuous welds could be achieved with up to 1 mm gaps and variable seam position with the developed control method. The acceptable weld quality could be maintained up to 0.6 mm gap in the employed autogenous welding configuration.

中文翻译：

---

通过灰度成像和摆动在机器人激光束焊接过程中进行焊缝跟踪和间隙桥接

随着系统可用性的提高和资本成本的降低，激光束焊接与机器人操纵器的使用正在扩展到更广泛的工业应用。传统上，激光焊接需要很高的定位和耦合精度。由于零件几何形状和定位的可变性，以及加工过程中可能发生的热变形，如果使用简单的夹具，接头位置和装配并不总是可接受或可预测的。这使得从虚拟 CAD/CAM 环境到真实生产现场的转变变得非常重要，从而限制了需要短零件准备的应用，例如小批量生产。需要使激光焊接操作适用于具有非严格公差的生产系列的解决方案，以服务于更广泛的工业应用。此类解决方案应该能够跟踪接缝并容忍待连接部件之间形成的可变间隙。在这项工作中，提出了基于具有外部照明的灰度同轴视觉系统和自适应摆动策略的机器人轨迹在线校正，作为提高制造工厂整体灵活性的手段。介绍了底层视觉算法和控制架构；还讨论了系统对不良照明条件和可变反射条件的鲁棒性。开发的解决方案采用两个控制循环：第一个能够改变机器人姿势以遵循不同的轨迹；第二个能够改变机器人的姿态以遵循不同的轨迹。第二，能够根据对接焊缝中形成的间隙来改变圆形摆动的幅度。使用复杂性更高的 AISI 301 不锈钢对接焊缝演示案例来测试该解决方案的功效。该系统在 2 毫米厚的平面不锈钢板上成功进行了测试，最大焊接速度为 25 毫米/秒，最大定位误差和偏航方向误差分别为 0.325 毫米和 4.5°。通过开发的控制方法，可以实现间隙高达 1 毫米和可变焊缝位置的连续焊接。在采用的自熔焊接配置中，间隙可达 0.6 毫米，可接受的焊接质量可以保持。