Can robots tackle the Amazon Picking Challenge?

The Amazon Picking Challenge, set by global online retailer Amazon, challenged roboticists to create a robot that can pick and stow 12 products from a shelf, into a bag. The Shadow Robot Company, designer and manufacturer of state-of-the-art anthropomorphic robot hands, looked to address the challenge with the help of precision bearing specialist, SMB Bearings.

Sounds simple, right? Well Amazon set this challenge because this is the only area of its warehousing process that isn’t currently mechanised, due to the sheer variety of objects to be packed. This presents a gap in robotics for a smart grasper that is able to recognise and pick different objects effectively.

Grasping is a problem for many industries, not just picking up an object – but grasping it well and ensuring that each grasp is correct for the object that’s being picked.

“More and more I’ve had people come up to me asking how we can improve robotic grasping,” says Rich Walker, managing director at The Shadow Robot Company. “The market is pushing us to improve the technology.”

Standard industrial grippers are built to pick up one thing, move it and put it down. If you have two different sets of objects, you’ll need two different grippers to perform the task. Modern manufacturers are increasingly diverse; producing reams of different products, often with custom finishes. Single-use tooling is no longer sufficient to keep up with demand. There is a need for tooling that is easily adaptable to pick up many different types of objects.

The robot engineers at the Shadow Robot Company have had plenty of experience manufacturing dextrous robotic hands for academia and nimble work, but this type of robot isn’t suitable for industrial applications. Instead, they looked to create a robust and reliable Smart Grasping System, with the intelligence to recognise many different objects and select the appropriate grasp. 

As part of the design process, the company needed specific bearings with very tight tolerances in dimensional accuracy. This is important for ease of installation and accuracy of the smart grasper’s movements. So, the team contacted SMB Bearings.

Precision bearings

Based on the requirements of this unique project, SMB Bearings recommended and supplied EZO thin-type precision bearings.

“The EZO thin-type precision bearings we provide have such accuracy as a result of EZO’s advanced manufacturing techniques and quality control,” explains Chris Johnson, managing director at SMB Bearings.
“If every bearing that comes off the production line is not the same shape and size every time, then the accuracy of the robot joint positions and overall behaviour would be compromised.” 

With the support of SMB Bearings and using a small sample of the EZO thin-type precision bearings, the research and development team at the Shadow Robot Company trialled different grease levels to find the optimal grease fill. As robotic movements need predictable and accurate some friction is essential to ensure this control. A process of testing and trial and error demonstrated the correct friction coefficient and its corresponding percentage grease fill for optimum performance.

SMB bearings then supplied the full amount of bearings required to this exact specification.

“These bearings were very easy to install in our Smart Grasping System,” states Luke Moss, robot engineer at the Shadow Robot Company. “This is a direct result of the quality, tolerance and consistency of these bearings, something that is essential for repeated assembly and high precision across multiple robots. 

Additionally, as the essential
core to the lifespan of any robot, we expect these bearings to outlive most other parts of the Smart Grasping System.”

Created to stand-out from standard graspers on the market, the Smart Grasping System is still in development, but ultimately will have built-in “smart” intelligence, allowing it to pick up many types of objects, reducing the need for numerous hands in a factory. Additionally, torque sensing on each joint will ensure the hand can make the most accurate and reliable grasp of whatever object it picks up.

“Industry 4.0 is on everyone’s lips at the minute, but we can’t forget the basics,” says Chris Johnson. “Bearings may not be considered smart, but if they enable the reliability and robustness of smart systems such as this Smart Grasping System, then I guess that is as close to being smart as a bearing can get.”

“I’m calling it transformative, because I know it will transform the manufacturing industry,” adds Rich Walker. “When companies have deployed robots in the past, they’ve used grippers that were designed to perform one function, so you get a robot that performs one function. The modern manufacturing industry needs more flexibility than this, and the Smart Grasping System will address this need.”

The new system will reduce the need for manufacturers to keep investing in new grippers for different tasks. Changing machinery incurs huge costs, and takes up a lot of time. Equally, having many robots performing one single task is wasteful in terms of both capital costs and floor space. Together, SMB Bearings and the Shadow Robot Company are working to solve this real-world manufacturing problem, freeing up floor space for a more efficient production line.

Taking on new challenges

Remember the Amazon Picking Challenge mentioned above? The Shadow Robot Company set its own picking challenge, using objects from Amazon’s challenge as well as others, to put the Smart Grasping System to the test. The system recognised and grasped these objects with ease, demonstrating the success of smart engineering and high precision bearings, to give the manufacturing industry what it desperately needs.