Delsys Prize 2003 Winning Proposal
Error Proofing Human Action in a Manufacturing EnvironmentKim Sherman, Senior Project Engineer, Sandalwood.
The current proposal is to use electromyography to monitor quality in automotive assembly plants. Providing hardware to monitor human movements throughout a work cycle can be used to ensure critical required work elements have been performed. The benefit of collecting data real-time while the movements are being performed will eliminate the need for non-value added quality checkpoints, additional equipment testing, and ensuring in station process control.
Falling in line with most others, one major automotive manufacturer has recently developed a machine monitoring system that enables quality of critical elements to be monitored during the assembly production cycle. This is technology that is easily implemented in manufacturing systems that provide stop station locations or systems that have built in buffers, but grows in complexity with large assembly lines containing moving conveyors in station. This system was developed to ensure compliance with current lean manufacturing objectives that emphasize the importance of building parts correctly the first time through a station. The goal is to provide a method of in station process control, verifying that parts are built correctly before getting passed down to the next workstation where a mis-build issue has potential to grow or continue on without being fixed. The current system records critical quality results such as fastener run down information or part selection verification data. The in station computers can record data and feed the results to other plant systems that control the conveyors and quality gate keeping stations. Currently the actions that can be monitored are limited to I/O devices. Actions that go unchecked can require extensive amounts of repair or worse get passed along to customers. Adding electromyographic I/O to an existing machine monitoring system would provide verification that required critical elements have been performed.
One specific application for the use of EMG is a task that requires the operator to engage an electrical connector along the firewall, behind the instrument panel. The operator must use both hands and apply approximately 20 lbs of force to ensure the clip has been fully engaged (see figure). Since this action is only required in 20% of the vehicles that travel through the station, one common error is that the task is not performed at all. The other error is that insufficient force is supplied to fully engage the clip. If this clip is not fully engaged the first time through significant diagnostic and repair work must be performed. The current proposal is to use EMG to first monitor gestures and ensure that the unique critical movement of reaching forward with both hands has been performed. The second task aspect that would be monitored is the level of force that was applied to engage the clip. Typically two operators will run these stations over the two work shifts. Practice cycles can be provided at the start of each workday and the period’s following rest breaks to allow for system calibration. This application of EMG applied to properly selected critical work elements can help to increase the quality of parts along with reducing costs associated with scrap, repairs, labor associated with non value added testing, testing machinery, and warranty issues. Some additional, more typical, benefits can be gained while collecting this real-time muscle activation data to name a few: using job templates to training new operators, monitoring the health status of high-risk individuals, and quantifying job requirements.