Artificial Intelligence (AI) Powered Robotic Each Picking for Order Fulfillment – Modern Materials Handling

Artificial Intelligence (AI) Powered Robotic Each Picking for Order Fulfillment – Modern Materials Handling

Robots have been around for a while now and some may now consider them a commodity.  They take inputs – product dimensions, pick and place locations – and produce outputs – finished pallets, unloaded or loaded machines. The inputs and outputs are controlled and consistent. What happens when the inputs and outputs can’t be controlled? Are you out of luck? Not anymore! 

Bastian Solutions has implemented robotic systems that allow robots to see, reason, and act to unknown inputs and produce desired outputs.

Leveraging AI Technology

These systems utilize artificial intelligence (AI) coupled with a vision system that can see various types of products in a homogeneous or heterogenous mix. The AI reasons and determines the optimal pick target for the robot. With the robot’s end of arm tooling (EoAT) configured into the artificial intelligence, it provides coordinates and angles for the robot to pick the product without colliding into external structures. Also, information is shared between multiple robotic systems, which lets them learn from each other for continued improvement. This also means that customers can introduce new products at any time with no robotic programming changes needed.

In these systems, we’ve achieved rates up to 1,400 picks per hour with 99 percent accuracy. This is largely dependent on the product, environmental conditions, and the robotic cell configuration but overall, these systems are very reliable and can meet or exceed the rates and accuracies delivered by humans. This brings an additional benefit with current COVID-19 social distancing requirements.

AI-powered robotic picking has been implemented in several industries, like food, apparel, pharmaceutical, ecommerce, parcel, manufacturing, and consumer goods. A few applications where these systems have been used are bin picking, decanting, return sorting, sorter loading, and several others.

Validating the Solution

At Bastian Solutions’ facility in St. Louis, Missouri, customers can see firsthand a robotic each picking cell that is setup for product handling demonstrations. Initially, two items are evaluated – vision suitability and robotic handling. For vision suitability, the product is put into the bin with a randomized assortment and the vision system can properly detect the products and provide targets to the robot.  For product handling, Bastian Solutions uses a variety of standard end of arm tools on the robot to show how the product can be handled.

After initial demonstrations and an order, detailed design begins and tackles the most challenging aspect of these systems: end of arm tooling. Bastian Solutions takes a step-by-step approach when designing their end of arm tools for each picking applications.

End of Arm Tool Design

To design an optimal end of arm tool, it’s important to look at not only the product itself but also the vision system interface and programming to ensure the system acts and learns in a way that supports your operations.

Products’ Dimensions and Shapes

  • Product dimensions and shapes dictate the method for handling the product. Is there enough surface area to use vacuum or does a compression gripper need to be used? Do multiple vacuum cup sizes need to be used to handle the minimum and maximum sized products? These questions are answered after product evaluation. 

Product Weight

  • Product weights determine the payload that a robot needs so that it can handle the heaviest product. The robot also comes with specifications for the end of arm tool and its center of gravity so that the robot can function properly. The robot’s end of arm tool must be less than the center of gravity specification for the selected robot while picking the various products. 

Product Material

  • The materials of the products play a large part into what components are selected for the end of arm tool.  For example, polybags need a soft material, like silicone, for vacuum cups to reduce the vacuum leakage. If compression is needed, the air pressure may need to be adjusted so that products with softer materials are not damaged.

Interface with Vision System

  • It is important to keep the end of arm tools “lean”. As an end of arm tool gets larger, it can impact the robot’s pick targets generated by the vision system.

Interface with Robotic Programming

  • End of arm tools need to be designed so that robots can reach the desired pick and place areas and avoid interferences based on a robot’s movements. 

Securing the Product During Robotic Movements

  • This is more of a challenge for vacuum end of arm tools. For high rate applications, the robot may need to move fast, and due to the products’ physical properties, it is important that the product is firmly held. To do this, multiple vacuum cups on individual vacuum zones may need to be used or an additional gripper may need to be added as a secondary gripping method.

Smarter Robotic Each Picking

After a proper end of arm tool is designed, robotic programming and artificial intelligence vision take over to move products to their destinations. Now those previously uncontrolled inputs and outputs are controlled for consistent, accurate robotic bin picking.

This technology has rapidly developed over the past couple years, and with the addition of artificial intelligence, the system is always learning and improving. With the impact of COVID-19 to meet increased customer demands and for employee safety, this technology is needed now. Bastian Solutions can help you implement, adapt it to your industry needs, and maximize the benefits of AI powered robotic picking for improved accuracy and throughput.

Source

Leave a Reply

%d bloggers like this: