How Robotics is Used by Bank of America, Amazon, NASA, Evergreen, Mamou-Mani Architects, and Obeta: Case Studies

Companies can develop innovative automation strategies using robotics, including those built with artificial intelligence (AI).

Across industries, businesses continue to look for ways to change, streamline, and improve their operations. The use of robotics, along with AI, for automation is helping companies achieve higher productivity and cost savings.

See below six case studies that show how organizations in different industries are using robotic technology:

1. Evergreen

Industry: Recycling and waste management

Robotics product: AMP Cortex robotics system

Challenge: Evergreen is one of the largest PET (Polyethylene terephthalate) recycling companies in the U.S., processing over 1 billion post-consumer PET bottles per year.

Before automating their plastic sorting process, Evergreen relied on employees to sort plastic pieces. However, according to Plant Manager Derek Guyer, “A human at their best can pick maybe 40 bottles a minute. We’re asking them to do that for 12 hours a day, standing in the same spot.”

Evergreen was also struggling to keep up with increasing volumes of bottles coming through their doors. With employees already strained, automation was a clear and necessary solution. Evergreen chose to partner with AMP Robotics to automate their sorting process using a combination of AI and robotics.

Outcomes: Evergreen had outstanding success with their automation strategy. AMP provided a team of pick-and-place robotic arms that were powered by an AI interface trained for object recognition. The AI can accurately and rapidly distinguish plastic bottles based on the material they are made of. This allowed Evergreen to increase their pick rate by 200%, bringing it up to 120 bottles picked per minute.

Employees commented that AMP’s interface was user-friendly and easy to program. This ease of use enabled Evergreen to be highly precise in how they programmed the AI-powered robots to separate bottles. As a result, they ended up with greater purity and consistency in their sorting process.

2. Bank of America

Industry: Finance

Robotics product: Pegasystems Robotic Process Automation

Challenge: As of 2022, Bank of America is the second-largest bank in the U.S., valued at $2.39 trillion. Bank of America serves millions of customers every day and employs over 200,000. They needed a way to modernize their operations throughout the front, middle, and back offices.

Bank of America wanted to be able to provide a better experience for customers, increasing efficiency without sacrificing quality. At the same time, employees needed ways to increase their productivity, but cumbersome manual tasks were simply taking up too much time. AI-powered robotic process automation was tapped as the best solution to these challenges, with tools supplied by Pegasystems.

Outcomes: Bank of America implemented their RPA strategy in stages over multiple years and maintains ongoing AI and robotics plans. In 2016, they brought in their first robot, and by 2019, they had 22 in use.

These robots performed a variety of tasks, which required “interpret[ing] applications for processes that involve manipulating data, executing transactions, triggering responses, and handling exceptions.”

The RPA tools that Pegasystems provided allowed Bank of America to integrate legacy systems, which simplified implementation since existing technology in the workplace didn’t need to change. At the same time, there was still plenty of flexibility and room for growth, which was important for Bank of America.

By integrating Pegasystems’s AI-powered robotic automation solutions, Bank of America was able to save hundreds of thousands of dollars and noticeably improve the consistency, accuracy, and efficiency of their automated banking processes.

3. Amazon Robotics

Industry: Warehousing and order fulfillment

Robotics product: Amazon Robotics and Amazon SageMaker machine learning

Challenge: Amazon is among the most famous companies in the world when it comes to automation. Their success stems from a combination of effective robots and machine learning.

They needed to automate their manual item scanning on the warehouse floor while also automating the hosting and management of the AI models that would run the robots. Amazon Web Services (AWS) SageMaker machine learning tools were the perfect solution.

Outcomes: Amazon Robotics had huge success with cloud-hosted machine learning for its AI-powered robots. AWS SageMaker was able to automatically scale models, adjusting size as needed without supervision. Using cloud-based hosting for their machine learning algorithms allowed Amazon Robotics to save almost 50% on inferencing costs.

The fleet of AI-powered warehouse robots delivered a 20% increase in productivity as well, boosting cost savings even further. This is one of a growing number of examples of AI resulting in tangible productivity improvements.

Experts have identified increased productivity as one of the top impacts of AI on the overall world economy. Amazon’s implementation in this AI-powered robotics case study demonstrates that businesses don’t necessarily need on-site computing infrastructure to tap into the power of AI in their robotics solutions.

4. Mamou-Mani Architects

Industry: Architecture and construction

Robotics product: Polibot

Challenge: Could the construction process be automated to deliver more resource-efficient architectural designs? This is the question that Mamou-Mani Architects set out to answer with their Polibot AI-powered robot. The project began with a UK competition to “use AI to create a machine that could transform construction sites.” Mamou-Mani Architects accomplished this with an AI-powered robot that uses machine learning to simplify the code required and increase functionality.

Outcomes: The Polibot is a marvel of engineering. It uses AI to rapidly generate architectural designs. Users can simply give the robot design criteria, and it will generate numerous design options and then analyze them to identify those with the most potential based on high-priority design criteria. The robot can then translate these digital designs into real structures.

Prototypes of the Polibot consist of pick-and-place mechanisms and winches suspended from overhead cables. The AI operates the robot, using cameras to lift, place, and align segments. It can incorporate 3D printing and CNC milling, as well.

The robot currently only works on a small scale of about four meters by four meters, although Mamou-Mani reports in this AI-powered robotics case study that the Polibot is highly scalable. One day, it could fully replace tower cranes and construction teams.

5. Obeta

Industry: Hardware supply wholesale

Robotics product: Covariant-Knapp Pick-It-Easy AI robot

Challenge: Obeta is a hardware wholesaler serving thousands of customers, with a warehouse that’s operational around the clock. They were struggling to maintain enough employees to keep things running smoothly and needed a solution for sorting, picking, and packing the thousands of orders they receive every day.

A basic pick-and-place robot would not do the trick — Obeta needed something that could distinguish between different items. An AI-powered robot from Covariant emerged as the ideal solution. It would combine automated pick-and-place with an effective and agile sorting AI algorithm.

Outcomes: Covariant supplied Obeta with the Knapp Pick-It-Easy robot, which uses AI to intelligently pick and sort objects. This robot was perfect for Obeta’s needs because it could integrate easily with their existing systems and accurately handle a variety of items, including items the algorithm has not seen before.

Obeta’s inventory changes frequently, so this capability was crucial. The robot’s performance peaked at an astonishing 600 items picked per hour with 99% accuracy, including new items.

6. NASA Jet Propulsion Laboratory

Industry: Aerospace

Robotics product: Boston Dynamics Spot robot

Challenge: NASA’s Jet Propulsion Laboratory wanted to develop an intelligent robot that could explore caves in dangerous environments, such as the surface of Mars. Traditional rovers on wheels are not ideal for the rugged and unpredictable terrain underground.

Additionally, cave exploration robots need to be fully autonomous, even in situations where they may lose contact with Earth. At the same time, the AI needed to be intelligent enough to observe and collect valuable findings from inside the caves.

So JPL partnered with Boston Dynamics to apply AI to the dog-like Spot robot. By combining JPL’s NeBula AI with the agility of the Spot robot, NASA was able to develop a new model for exploring harsh environments. The technology could be applied on Earth in fields like mining, construction, or even the utilities and power industries.

Outcomes: The collaboration between NASA and Boston Dynamics led to a successful demonstration of how an AI-powered robot could explore Martian caves searching for signs of life. As the Spot robots moved through a cave here on Earth, they were able to map it in real time and send data back to a base station.

Sensors and cameras on board the robots were used to analyze the environment underground, scanning for biosigns the AI model was trained to recognize. The agility of the Boston Dynamics robots made a monumental impact on the success of the project.

“What’s so exciting about Spot is how flexible it is and how maneuverable it is,” says Jennifer Blank, a NASA lead scientist on the project.

“I can envision scenarios where different Spots have their own assigned roles and their own assigned specialties.”