Vision systems will drive the next major wave in industrial robotics

  • New generation of industrial robots will cope with an unstructured environment
  • Robots will be able to ‘see’ the shape and orientation of items
  • 3D vision system combines camera with scanning laser 
  • Key application for vision systems is quality control

Whilst the majority of current robots operate in a highly structured environment where the workpiece and workstation are precisely known, the new generation of industrial robots are able to cope with an increasingly unstructured environment where workpieces can be different one-to-another or they can be presented in variable positions such as on a leaning and twisted pallet stack.

Sydney-based company, Applied Robotics, has been building special purpose automation and robotics machine systems for Australian and export markets since 1985, with over 500 systems now installed worldwide.

The company’s managing director, Dr Paul Wong, believes that extending the robot’s capability to embrace these less-structured areas of manufacturing environments is the next major wave of change.
"The next wave involves equipping the robot with sensing capability, which in turn enables flexible infeed to the robot because it can ‘see’ the shape and orientation of items.

 

This illustrates a typical brick with a crack identified by the Vision System.

"The best sensor candidates for the job are vision systems whereby a camera conveys a digital image to the robot for analysis and extraction of shape, size, position and orientation data of the workpiece to be picked. The vision system’s camera(s) can be mounted on the robot itself or fitted as an eye over the work space," Wong told PACE magazine.

"At a more sophisticated level, a 3D vision system which combines a camera with a scanning laser can produce a 3D map of the scene, giving the robot an extra dimension of data with which to translate a more complex scene.

Some sectors did not have capable automation technologies available to them

Wong elaborates: "Some manufacturing sectors have easier tasks to perform than others, so some, including the automotive and electronics industries, were able to adopt automation technologies earlier than others. Other sectors, such as the footwear and clothing industries utilising porous and floppy materials that required sewing, did not have capable automation technologies available to them, and as a result they were lost to the cheap labour countries.

"Now 20 years later, a raft of emerging new technologies, including high level vision systems, will be able to help these sectors. Currently, less than 10 per cent of manufacturing companies would have a vision system, but in 10 years it is likely that only 10 per cent will not have this technology."

A key application for vision systems in manufacturing industry is quality control. In the last five years, Applied Robotics has installed vision system technologies from Omron to perform bricks and blocks quality control functions as a part of larger automation lines for bricks and blocks manufacturers.

 

A typical robot palletising blocks in a complex pattern.

Wong says that the vision technology is usually part of a multi-robot system to de-hack layers of fired bricks off kiln car stacks and singulate them into a line-astern column. "This line is then conveyed through a vision system module for on-the-fly quality inspection to separate any rejects to waste, collate the passed bricks and blocks into arrays for robot pickup, and palletise into customer pallets," he said.

"Within the quality control module, the multiple camera vision system looks at all four sides of the brick, searching for faults such as cracks down to 1mm wide, pit-holes, scars, and edge and corner chipping. Depending on the type of fault, the brick can be rejected for that fault alone or intelligence is applied to group faults in an additive manner.

"For example, pitting on a non-critical face is acceptable unless there are many pits in a cluster, or if more than one of the non-critical faces on one brick are pitted. General brick parameters such as colour, overall dimensions, a curved brick, a wedged brick, or a waisted brick are also monitored and rejected if out of tolerance.

"Use of vision system based quality control has enabled the products to be checked thoroughly and quickly at rates up to 60 bricks per minute, ensuring the delivery of only good products to customers."

Vision system performs product identification and parameter set-up

At Dux Hot Water in the Southern Highlands of NSW, Applied Robotics has installed vision systems, again utilising Omron technology, to perform product identification and parameter set-up for a flexible multi-robot assembly and welding workcell wherein the infeed of workpieces is unstructured.

Wong explains that while the workcell has been designed to assemble and weld sockets to dome tops and flanges to dome bottoms, for six sizes (diameters and heights) of domes, the infeed of the domes is in a stream of mixed types and in a mix of the six different sizes.

"These dome tops and dome bottoms form the end caps for hot water cylinders. Each arriving dome in the robot cell is simply arrested and clamped into a common reference corner," he said.

"Here, a vision system is employed to identify the dome type (small hole for socket or large hole for flange assembly), and measure and report the socket or flange hole centre position and the hole height on the as-clamped dome. This information is then given to an assembly robot to pick either a socket or a flange from its magazine and assemble the picked item into the hole in the clamped dome.

"Then, the same measured data is sent to a welding robot to weld the socket or flange when the assembly robot is withdrawn. The welding robot tracks the weld-line at the appropriate diameter around the as-measured hole centre and at the as-measured height.

"This system enables a mix flow to be processed. It is ‘jig-less’ and therefore has a zero set-up time for product change. A traditional solution would require batches of the same domes to be run, requiring specific tooling and set-up time by skilled operatives for each batch." Chris Burgess, engineering manager at Dux Hot Water, says two vision systems are used in the dome welding cell.

"The first vision system determines what type of dome is being processed and signals the loader robot to select the appropriate components to be welded to the raw dome. The second vision system accurately measures the actual location of the various features on the dome," he said.

"This information is first passed to the loader robot to enable it to accurately position the components, and then passed to the welder robot to enable it to weld the components to the dome. Previously, domes were hand loaded into various welding jigs. Two of the key goals of the dome welding cell were to reduce manual handling risks and the proximity of personnel to welding operations.

"The successful implementation of vision systems has achieved this along with improvements in labour productivity as staff are no longer required to carry out non value adding movements."