High precision visual servoing control of robotic manufacturing systems

在制造业中，为了降低劳动力成本和提高生产能力，对自动化机器人制造系统的需求越来越大，例如用于管道和部件对齐的机器人装配、用于取箱的系统以及用于钻孔和焊接的紧固系统，这些系统要求高精度和快速的机器人定位。这些苛刻的要求对机器人在非结构环境中处理大尺寸、高有效载荷的工件提出了挑战。将机器视觉系统集成到机器人中可以应对这一挑战，是成功实施机器人制造系统的关键一步。该视觉系统模仿人类的视觉，并允许对环境进行非接触测量。当它被用作机器人传感器时，视觉系统可以补偿机器人的定位误差，这种基于视觉的机器人系统被认为在提供制造精度和灵活性方面是有效的。然而，由于工业环境中光照的变化、物体的变化以及机器人固有的非线性动力学，很难保证机器人制造系统的定位精度。虽然大量的学术研究和工业发展都致力于视觉伺服，但如何在未知环境下实现大尺寸零件的高精度和快速定位，仍然是学术界和制造业都在寻求的高性价比解决方案。这项提议的目标是建立可靠和实用的集成解决方案，使机器人制造系统能够在工业环境中实现高位置精度。该系统对光照变化、标定误差等不确定性具有较强的鲁棒性，并能处理机器人的非线性。最终的目标是开发性价比高的视觉伺服系统，使目前的机器人制造系统具有视觉上的高端。

To reduce labor cost and increase throughput in the manufacturing industry, there is an increasing demand for automatic robotic manufacturing systems such as robotic assembly for piping and parts alignment, bin picking systems, and fastening systems for drilling and welding, which require high precision and fast robot positioning. These demanding requirements pose challenges for the robots to handle the large-scale workpiece with high payload in an un-structural environment. The integration of the machine vision system into robots can address the challenge and is a key step towards the successful implementation of robotized manufacturing systems. The vision system mimics the human sense of vision and allows for non-contact measurement of the environment. When it is used as robotic sensors, the vision system can compensate for the robotic positioning errors and such vision-based robotic system is known to be effective in providing precision and flexibility of manufacturing. However, due to changing illumination in industrial environments, variations in objects and the inherent nonlinear dynamics of the robots, it is very difficult to guarantee the positioning precision of the robotic manufacturing systems. Although a lot of academic research and industrial development have been devoted to visual servoing, cost efficient solutions to high precision and fast positioning of large-scale workpiece in an unknown environment are still sought by both academia and manufacturing industries. The objective of this proposal is to establish reliable and practical integration solutions that will enable robotic manufacturing systems to achieve high position accuracy in an industrial environment. Such systems are robust to the uncertainties such as changing illumination and calibration error and can deal with the nonlinearities of robots. The ultimate goal is to develop cost effective visual servoing systems to render the current robotic manufacturing systems high end ones with the sense of vision.