ARRA: Machine-Human Cooperative Control of Welding Process

Grants and Contracts Details

Description

For many critical applications, a welder's experiences and skills are crucial in producing quality welds meeting specifications. However, such skills require a long time to develop and "United States is in the midst of a welder shortage that is expected to intensify". Using a machine to cooperate currently welder-controlled process leads to a machine-human cooperative control of welding process with a welder in loop. An innovative way is thus provided to enhance welder capability without the needs to modify or change the current equipment in use. This project aims at establishing the foundation to provide the science, methodology, and knowledge base for this new kind of control. The combination of human intelligence and vt!rsatility with continuous and accurate machine supervision and adjustment, a distinctive characteristic of this proposed machine-human cooperative control, can potentially become an effective way of thinking to invent next generation welding and other manufacturing machines. Intellectual Merit: The intellectual merit of the proposed research primarily lies in (a) the sensory helmet idea that innovatively takes advantage of welder natural movement to facilitate an automatic process monitoring from an optimal position; (b) critical thinking and analysis of current welding practices for the innovative concept of cooperative control; and (c) innovative methodology for adaptive modeling using accelerometer based welder action monitoring. In addition to these innovative ideas/concepts/thinking, the intellectual merit is further enhanced by the nature of the research that is challenging, fundamental, and paradigm shifting. In particular, our proposed modeling of human welder reaction process and welder-operated process with unique operational characteristics are fundamentally different from industrial physical processes. Our prediction and control of welder-controlled process innovatively include a human welder in the loop, whose behavior is characterized and modeled for model based predictive control. We expect to demonstrate all these principles through experiments that require all resultant solutions to be robust and realistic. Further, in-depth understanding will be needed and developed in areas including welding process, welder behavior, linear and non-linear process modeling, advanced control, image processing, ergonomics, etc. Broader Impact: Welding is typically the last operation during fabrication of high value added product. Precision joining is thus a critical capability that US must maintain through either automation or stabilization of qualified labor force. The proposed cooperative control that uses machines to assist human welders and is capable of recording on-line for welder training provides an effective and innovative way to help stabilize this labor force while enhancing/improving their capability, durability, and health. It will establish the foundation for a new kind of next generation welding machines and may also potentially inspirit inventions in other manufacturing/industrial sectors. The multi-disciplinary nature of the proposed research and collaboration with industry will provide opportunities to train next generation academic and manufacturing experts/researchers in a wide range of levels from high school, undergraduate, through PhD in forms of hands-on design course, research experience, exploratory thesis/dissertation research, application-oriented case study, knowledge dissemination and possible commercialization with industry partners. It will also help the formation of a highly capable team (Zhang and Yang) to target and resolve other highly challenging issues toward better next generation welding and manufacturing machines.
StatusFinished
Effective start/end date10/1/099/30/13

Funding

  • National Science Foundation: $400,000.00

Fingerprint

Explore the research topics touched on by this project. These labels are generated based on the underlying awards/grants. Together they form a unique fingerprint.