A bin arrives at a robotic cell and it’s filled with parts of different sizes that have just been tossed in. The disorganized, tangled, and overlapping mess presents a challenge. Some of the pieces needing extraction weigh 30-pounds but become like a 50-pound or 60-pound part because of the other items stacked on top.

A robot swings into action with its software sending commands and its vision system processing images faster than the blink of an eye. The items are sorted without any problems and are ready for the next step.

Random Bin Picking Robot
Random Bin Picking Robot (Source: http://www.a3automate.org)

Researchers and robot manufacturers have been dreaming of this scenario for many years. Simply buying one robot that comes with all the components needed to pick and sort completely random objects isn’t yet a reality. The quest to perform random bin picking with ease has led to tremendous breakthroughs, though, having a wide range of benefits and uses.

Industrial robots have moved from being large and stationary without any perception to being flexible and equipped with sight and sensitive touch. The flexible robot has become a cost-effective alternative, allowing large and small companies to use it for multiple tasks and handle completely different product lines.

What’s Needed to Pick and Sort

Why is random bin picking so important? Unloading parts for a machine or fulfilling orders are tedious tasks and put workers at risk for injury. Getting robots to function well with this type of task expands their capabilities and helps companies of all sizes and in many different industries operate with even greater productivity.

Random bin picking requires “a convergence of technologies, particularly three main components that raise the robot’s intelligence: sensors, software, and end-of-arm tooling,” as noted in the article Robotic Bin Picking – the Holy Grail in Sight. For a robot, an unordered bin is chaotic in a micro-environment that has no structure and no predictability. Operating in a “Wild West” environment requires flexibility.

Improving Upon People

Developing the technology also shows how robot development mimics what people do well, like deciding what items to pick up and how to arrange them. We have brains. They have some software. We have sight and depth perception. Robots have 2D and 3D orientation and sometimes utilize both as described in the article on visiononline.org, 2D or 3D Machine Vision? Why Not Both?

Our arms and hands are perfect for lifting and gripping, although most of us can only lift light loads. Robots have end-of-arm tooling that can be fitted with grippers for different tasks. Usually, only a small amount of downtime is needed to change out the grippers.

Perception itself is a major advance as noted in Intelligent Robots: A Feast for the Senses. Sensors are an important reason that robots have sight and the components are becoming less expensive. It’s an area that’s still open for development and growth.

Coming up with a solution for picking random parts from a bin is a creative challenge. Motion Controls Robotics of Fremont, Ohio, is an integrator and has developed a robotic cell that uses a 3D area scanner to locate parts, a magnetic end of arm tooling to sort, and a two-finger gripper to pick up the parts.

Achieving the end result wasn’t easy. Motion Control Robotics had to try different vision systems and grippers before settling on the final configuration. The robot is shown operating in a video titled Robotic Bin Picking Cell and uses a sensor that tells the robot if a part was placed properly in the tray. Random bin picking is on its way to becoming so routine that someday we might take it for granted.

*This content was reposted in part from Association for Advancing Automation, May 11, 2016.

>Read this article in full.

Breakthroughs in the Quest for Random Bin Picking