Urgent need for automating bin picking
Nearly 40% of the manufacturing labor force in the U.S. moves parts between bins and manufacturing machines. As the country is enjoying its lowest unemployment in nearly 50 years, the need for automated bin picking is urgent. This is also true for other countries suffering from a labor shortage due to the mass-retirement of baby boomers and other factors.
Despite the urgent need, the adoption rate of utilizing an automated bin picking system remains low. Bin picking is not especially complex when the objects to be picked are well organized. When the parts are situated in an organized pattern, they can easily be identified and picked. This task can be automated using 2D Vision, imaging, and analysis, however, its complexity increases significantly whenever the objects are randomly placed. Due to its complexity, random bin picking has yet to be perfected by automation systems. Picking randomly positioned parts from a bin and placing them precisely into a machine is simple for a human worker, but a very difficult task for robots. This is because the robots need to be able to grasp randomly placed parts as well as reaching deep into the corners of a bin without colliding with the bin or the other parts inside the bin.
A number of different technologies are beginning to overcome some of the challenges in robotic bin picking, however, including advanced vision systems, improved sensors as well as software and gripping technologies. Thanks to the advances in machine vision, 3D sensors and AI, some experts predict that robotic random bin picking will finally become mainstream this year.
Bin picking with advanced vision systems and AI
Vision systems help robots to adapt to the environments around them as well as to locate objects for picking. Advanced vision systems give robots the flexibility they need for picking randomly placed parts in a constantly changing environment. To provide a robot with accurate coordinates for each movement often requires the use of multiple cameras (stereo vision, overhead, and end-of-arm mounted cameras, etc.). Getting the right coordinates when the robots and camera(s) move around and constantly changing the lighting and the field of view can be a challenge, but new algorithms and greater computing power have been helping robot engineers to address this challenge.
For example, FANUC America’s picking robot equipped with its iRVision uses a “bin picking” wrapper, projects light strips into the bin and takes 16 pictures quickly to guide the robot to the part. The advanced algorithms also help picking robots avoid making contact with the bin. Meanwhile, Adept Technology’s SmartVision MX Vision Processor uses up to eight cameras for robot guidance applications.
AI also helps the challenges of automated bin picking. The conventional process of training bin picking robots is to teach them many rules so that the robot knows not only which parts to pick up, but also how to pick them up. This is a time-consuming and cumbersome process, which involves a lot of trial and error. Human operators need to tell the robot what and what not to do whenever it makes an error in order to refine the process. To address this challenge, Fanuc has introduced an AI-based tool, which simplifies the training process. With this tool, a human operator only needs to look at a photo of some randomly placed parts in a bin on a screen and tap a few examples of what needs to be picked up. It is quite difficult for the human operator to show the robot how to move parts in the same way that an operator can, but, thanks to AI technology, the operator can now teach the robot more intuitively, like the way we show a small child how to sort their toys. Fanuc’s tool can also be used to train several robots at once.
Today’s bin picking technology dramatically differs from the technology available a few years ago. Recently, TM Robotics and Toshiba Machine launched the TSVision3D system, which incorporates two integrated, high-speed stereo cameras capable of 30 frames per second for continuous, real-time 3D images. The cameras manage image capture, processing and parallax operations to identify the items’ positions, and the vision software adds easy model registration without requiring complex CAD data.
New players with new solutions
Several startup companies are also offering all-in-one bin picking systems based on new technologies. Their systems can work with many types of machines and commercially available 3D cameras. CapSen PiC (pick in clutter), developed by CapSen Robotics, automates the physically demanding and repetitive task of bin picking, enabling robots to pick objects from a cluttered bin and to place the objects in a preset pattern. FlexiPick by Bluewrist also enables a robot to pick up parts randomly placed in a bin and present them to an assembly line. Another new player in the bin picking marketplace is a Slovakian company called Photoneo.
Photoneo has developed an all in one picking system that works with several different robot brands, including Fanuc, Kawasaki, Omron, Universal Robots and Yaskawa. Photoneo’s Bin Picking Studio released in 2018 includes all of the components needed for a bin picking application like an automated robot-scanner calibration, CAD-based or AI-based object recognition, robot path planning, a collision avoidance system and a 3D sensor among others. Its 3D vision system can quickly recognize randomly placed parts in big scanning volumes with high resolution and high accuracy. In addition, its new PhoXi 3D camera features a carbon fiber body and can acquire 1,068 by 800 point clouds, plus texture at up to 60 frames per second with active ambient light rejection. Photoneo also offers the PhoXi 3D scanner that can scan an entire pallet or box at once. Five models are available and can cover a variety of scanning volumes for applications ranging from small PCBs to big boxes used in automotive production. Photoneo’s new MotionCam-3D, which the company claims has the highest resolution and the highest accuracy area-based 3D camera for scanning in rapid motion in the world, was the winner of the coveted VISION Award 2018 at VISION, the world’s leading trade fair for image processing which takes place every two years in Stuttgart, Germany and was ranked among the inVISION Top Innovations 2019.
The way forward
With better vision systems, cameras and greater computing power, bin picking technology has improved significantly as the accuracy, resolution and performance of sensors have improved. At the same time, prices have decreased, software has been simplified and processing speeds have increased. Today’s systems are also easier to set up and configure, and they’re more flexible to adapt to changing requirements. The ease of setting up and greater flexibility will have great benefits given the fact most bins and manufacturing machines are found at SMEs which may not have the financial and human resources and expertise efforts needed to create a bin picking system.