Robotic solution achieves OH&S improvements

Luxfer Gas Cylinders in Sydney's west recently installed a robotic solution to automate its powdercoat paintline. The gas cylinder manufacturers wanted to improve efficiencies at their manufacturing plant and reduce the need for manual lifting by their employees.The powdercoating process is a labour intensive operation requiring the gas cylinders - weighing up to 43 kg - to be manually lifted and hooked onto a moving overhead transport chain at a rate of up to four per minute. After painting and baking, the hot cylinders are unhooked and loaded onto a downstream conveyor for further processing.

"When the system went live at our plant, it not only significantly reduced the risk of back injuries and RSI amongst our operators but we also achieved productivity improvements up to 33%," states Geoff Tan, engineering manager at Luxfer. "When our OH&S task force identified the task of hanging and unhanging cylinders at the paint line moving conveyor as being a high risk task, we searched globally for an automated solution. Nothing was available to meet our needs," explains Tan.

Luxfer's requirements were difficult because most of the world's robotic applications in industry operate on the premise of 'dead-reckoning' - that is, the robot assumes that the workpieces it manipulates are correct and in the right place every time. In an environment where a job is 'unstructured' or if the product itself is variable, human beings still perform better, even when the loads carried are heavy and/or awkward and the job is dull and repetitive. Luxfer engineers along with Applied Robotics reviewed the process and developed a world first automated process using an Omron machine vision system for real-time sensing and two ABB 6400 series robots to handle the cylinders.

The machine vision system senses the relative positions of the cylinder and the paintline hook in real time. The positions of both items are communicated to the robot to effect a successful hooking on. The sensing process involves the robot 'floating' the cylinder a nominal 100 mm distance below the empty paintline hook and moving with it in synchronism. The vision system, which uses two cameras, acquires a 3D view of the moving hook and cylinder and calculates the positional error in x/y/z coordinates. This error is then communicated to the robot, which acts to correct it to engage the cylinder onto the paintline hook. The vision sensing and hooking on part of the cycle takes about 1.5 s. Geoff Tan is very pleased with the results the system is producing: "The system was installed and commissioned about a year ago. It has been working extremely well, consistently hooking all sizes of cylinders despite the large variations in both the cylinder shape and the paintline hooks."