Search
Menu
Edmund Optics - Manufacturing Services 8/24 LB

Vision-Aided Robotic Grippers Automate Produce Packing

Facebook X LinkedIn Email
Loaded with vitamins and minerals, the sweet potato is a popular, nutritious, and versatile food — whether simply roasted or prepared in a fancy casserole. Sweet potatoes are also a profitable crop. But as with most other businesses, exporters of the starchy vegetable are not impervious to labor shortages and inflation.


 
  Soft Robotics’ mGrip system enabled Assatec Robotics to automate a conveyor-based process for picking and packing sweet potatoes. The mGrip’s hollow gripper fingers are pneumatically controllable at low pressure, allowing the grippers to gently grasp objects without the need for an operator to adjust the variables of the overall automation system. With the grippers installed, the system avoided process stoppages. The addition of a software component enabled proper placement of the sweet potatoes into packing boxes of the right size. Courtesy of Soft Robotics.

To ensure that sweet potatoes remain on menus and in kitchens worldwide, companies can deploy advanced technologies, such as robots and machine vision, to automate processes and thereby increase productivity, cut costs, and drive revenue. Challenges exist, but the right combination of hardware and software can help to ensure that production continues.

At a time of global labor shortages and high food costs, creative solutions are becoming increasingly necessary. Recently published data from the Association for Advancing Automation (A3) indicates that robot sales hit record highs in North America for three straight quarters that spanned 2021 and 2022, according to the organization. Robot use in the food and consumer goods industry continues to climb, both year over year and quarter to quarter.

3D cameras above the boxing area capture 3D images of each box as it is packed (top). The software analyzes the images and decides where the next potato is to be positioned in the box. Once the robot places a potato, the system immediately begins the process of picking another potato and finding a place for it in the box (bottom). The software-supported imaging system then captures a 3D image of the fully packed boxes to ensure that no potatoes are sticking up above the top edge. Courtesy of Soft Robotics.
3D cameras above the boxing area capture 3D images of each box as it is packed (top). The software analyzes the images and decides where the next potato is to be positioned in the box. Once the robot places a potato, the system immediately begins the process of picking another potato and finding a place for it in the box (bottom). The software-supported imaging system then captures a 3D image of the fully packed boxes to ensure that no potatoes are sticking up above the top edge. Courtesy of Soft Robotics.

 
  3D cameras above the boxing area capture 3D images of each box as it is packed (top). The software analyzes the images and decides where the next potato is to be positioned in the box. Once the robot places a potato, the system immediately begins the process of picking another potato and finding a place for it in the box (bottom). The software-supported imaging system then captures a 3D image of the fully packed boxes to ensure that no potatoes are sticking up above the top edge. Courtesy of Soft Robotics.

According to Research and Markets, the global food robotics market was valued at more than $2 billion in 2020 and is estimated to reach $5.8 billion by 2031, registering a compound annual growth rate of 10.4%.

Finding a soft touch

Although automated systems can perform grading and sorting tasks, many require human counterparts to either guide the system or perform additional steps — even within a single process.

PI Physik Instrumente - Microscope Stages MR ROS 11/24

At a major sweet potato exporter in Israel, a machine sorted potatoes by size. Potatoes of various sizes traveled from different sections of a main conveyor belt. At conveyor output stations, employees then packed the potatoes neatly into boxes. Each station required a minimum of four employees to pack the sweet potatoes.

The company wanted to automate the process, so it hired automation systems company Assatec Robotics.

Even manually, picking sweet potatoes in a high-speed process is challenging because of the variety of shapes and sizes. Assatec sought out robotic grippers capable of handling the varying sweet potatoes. When other grippers on the market proved inadequate, Assatec turned to Soft Robotics’ mGrip system. The company’s IP69K-rated gripper uses an advanced materials science-based approach to safely and efficiently pick up products of all types. The mGrip’s patented elastomeric plastic grippers are made with food-safe materials. Hollow gripper fingers with flexible holds are pneumatically controllable at low pressure, allowing the grippers to gently grasp objects without the need for an operator to adjust the variables of the overall automation system.

After the Soft Robotics gripper was integrated into the system, improvements were immediately noticeable.

“With the previous system, the gripper could hold parts that varied in shape, but we found limitations,” said Assatec’s CEO, Or Levy. “With the Soft Robotics gripper, however, we could even handle tea cakes without breaking them.” Without the gripper, he said, the application would not have been possible.

3D vision-guided pick and place

In the automated process, sweet potatoes travel down the conveyor belt in a random array and pass under an NSIX CVK5 3D camera positioned above the input conveyor. Based on Intel RealSense technology, the camera — which offers 1280 × 720 depth resolution at up to 30 fps at maximum resolution — captures 3D images of the potatoes. NSIX Vision Keys software running on an Intel Core i7-9700 industrial PC with an NVIDIA GeForce GTX 1650 graphics processing unit guides a FANUC M-3iA/6S delta robot equipped with soft grippers to make picks.

For potato placement optimization, the collaborators developed a custom volume-fitting algorithm within the Vision Keys software. Two additional CVK5 3D cameras are positioned above the boxing area, where they capture 3D images of each box being packed. The software analyzes the images, considering each potato’s size and shape, and finds a place for the potatoes in the box. Once the robot places a potato into a box, the system immediately begins the process of picking another potato and placing it into the box.

Courtesy of iStock.com/rebeccafondren.

 
  Courtesy of iStock.com/rebeccafondren.

'With the Soft Robotics gripper, we could even handle tea cakes without breaking them.'
 — Or Levy, CEO of Assatec Robotics 
Once the box is fully packed, the system captures another 3D image to ensure that no potatoes stick up above the top edge of the box. If any potatoes stick up, a vibratory mechanism under the box helps spread the products out to even them — another job that was previously performed by employees.

Options in automation

In an increasingly automated world, produce companies also need flexibility. Assatec therefore designed the system with versatility in mind. According to Levy, the system can work with any type of fruit or vegetable.

With Assatec’s algorithms, the Soft Robotics gripper, and 3D vision, the system opens up new possibilities in automation, he said.

News editor Jake Saltzman contributed to this piece.

Published: November 2022
Glossary
picking
The process whereby pitch is broken and the lenses or blanks are removed from the block.
deep learning
Deep learning is a subset of machine learning that involves the use of artificial neural networks to model and solve complex problems. The term "deep" in deep learning refers to the use of deep neural networks, which are neural networks with multiple layers (deep architectures). These networks, often called deep neural networks or deep neural architectures, have the ability to automatically learn hierarchical representations of data. Key concepts and components of deep learning include: ...
Vision in Action3D inspection3D imagingcamerasSoftwaresoft roboticsroboticspickingAIdeep learningcomputeautomationgrippersindustrial

We use cookies to improve user experience and analyze our website traffic as stated in our Privacy Policy. By using this website, you agree to the use of cookies unless you have disabled them.