Vision, Software Enhancements Advance Robots
- Sunday, 01 June 2008
Wodoslawskyl refers to the software-driven technology as “force control”, which ABB implements in robots for materials finishing and assembly, such as assembling an automatic transmission. “Before robots could be equipped with force control, properly installing a torque converter onto the transmission spline gear required a person to place the converter onto the shaft, rotate it slightly left and right feeling for splines, then push the converter onto the assembly. A robot using force control not only eliminates an ergonomically undesirable task, but greatly improves quality by reducing the risk of damaging gears during assembly.”
SmartTCP Inc.(Farmington Hills, MI) has developed a robotic welding system that relies on hardware and complex software to automate small batch welding applications. “The way the market does it is to teach the robot to search for specific parts in certain instances,” said Effi Lebel, founder and CEO. “We developed searching technology that is good for any part, with any geometry, in any position.”
SmartTCP’S software analyzes the 3D CAD file of a structure’s final assembly drawing as though it were through the eyes of an expert welder. It will then implement the most sophisticated welding and robot technologies. “Because we have a 3D model of the part, we have a virtual image of the part. We can immediately recognize the part and implement the right type of search.”
Advanced vision systems have raised the level of robot intelligence, particularly as manufacturers integrate them with the robot rather than design separate systems. “With latest controllers, we’ve been able to move machine vision processing into the same CPU as the robot,” said Fanuc’s Dinsmoor. “Every robot can now incorporate a camera and cable to implement machine vision. Every robot can now see without the addition of costly equipment.”
ABB’s vision capability enables the robot to see in three dimensions, using a single camera. “In the past, multiple cameras or a camera and a laser were required to locate an object,” said Wodoslawskyl. ABB teams the single camera with highly sophisticated software to find, pick up, and manipulate objects. This approach reduces the robot’s susceptibility to jars and jolts typical in a manufacturing environment, and speeds system calibration.
According to Wodoslawskl, automotive companies are using vision technology to assemble automotive engine heads onto the engine block. “Because the system can find the engine and retrieve it, then pick it up and precisely place it on the engine block, a process that was once manual is now automated. This reduces the change of damage to the head and the engine block, and removes a backbreaking task for the auto worker.”
Robot vision systems now also can compensate for lighting variations in industrial environments. “Ten to 15 years ago, a robot could find a part on a conveyor only with appropriate lighting,” according to Bob Rochelle, North American sales manager for Kawasaki Robotics (USA) Inc. (Wixom, MI).
Improvements in electromechanical components are also enabling robots to handle higher payloads. “Maximum payloads used to be around 200 kilograms, now they’re 500 to 750 kilograms,” said Rochelle. He added that Kawasaki’s robots are used by automotive manufacturers to lift entire car bodies weighing 800 to 900 lb.
To speed productivity, robots are also incorporating flexible tooling that eliminates the need to stop the robot to initiate a new assembly sequence. “As the part comes in, one can switch the system and the robots and tools move,” added Rochelle. “The tooling is adjustable, with the actuator sliding left and right.”
As robots become more intelligent, they will incorporate some of the safety provisions previously implemented on assembly lines, according to Fanuc’s Dinsmoor. These provisions ensure the robot works within the programmed range of movements.
“We’ve started to introduce software safety features within the robot - particularly for ones with larger reach. People want to keep down the work cell size if possible, so we’re implementing functions within software to constrain the robot.”
CoroWare’s Mandrel added, “We’re definitely seeing more situational awareness. Whereas previously a robotics algorithm was used to control one arm, now it is to control two arms and prevent them from hitting one another. “
Lloyd envisions a future with robotic assembly cells comprising multiple, smaller robots working together. “There will be layers of decision making and connections,” Lloyd said. “You can have balanced workloads, with different processors handling different (robot) functions.”