How Waste Robotics, Greyparrot are enhancing sorting robots

Greyparrot and Waste Robotics partnered earlier this year to enhance Waste Robotics’ sortation capabilities. The company creates robots that sort a variety of wastes, including construction and demolition, recyclables, metals, and municipal solid waste, and the everyday items tossed out by the public.

The company uses FANUC robot arms paired with its proprietary AI and gripper technology. Now, it’s also using Greyparrot’s AI to ensure its robots are being used at the most important parts of the waste sortation process. 

Waste Robotics creates what it calls physical AI that helps robots identify and pick up a variety of items. The added AI layer from Greyparrot uses cameras to track all the materials passing on conveyor belts and create real-time insights on a live dashboard.

With Waste Robotics, the Greyparrot Analyzer is at work characterizing material streams and identifying the most impactful opportunities for automated robotic sorting. The combined product called the Robot Validator, allows Waste Robotics’ customers to know that they will be getting the most out of their robots from day one.

The Robot Report connected with Eric Camirand, CEO of Waste Robotics, and Gaspard Duthilleul, COO of Greyparrot, to learn more about the partnership.

Tackling the “last mile” of waste sorting

Waste sorting is an ideal use case for robotics. It’s repetitive, back-breaking work that involves handling a variety of materials. It’s no surprise many material recovery facilities (MRFs) have a difficult time finding enough staff. Still, despite these challenges, MRFs may be hesitant to try automation because of the sheer variability in their facilities.

When it comes to delivering goods, the “last mile” is a term used to describe the final, most difficult part of the delivery process, when the goods are handed off to the customer. Camirand said that waste sorting is dealing with a similar problem right now. 

“When we talk about waste sorting, the last mile is often what we call quality control,” Camirand told The Robot Report. “So there are different technologies out there that can sort, but when you are in the final finicky few picks, the critical picks to be done to improve the quality of the product or service level, it’s hard to deploy a lot of equipment like optical sorters or other types of machines to do this. So robots kind of come and really replicate what humans are doing. Right now, the most effective way to remove these last few contaminants in, for example, the plastic line, is really with robotics.”

Waste Robotics said it specializes in this “last mile” of waste sorting. The company has a variety of gripper types, from suction grippers to multi-fingered grippers, that enable it to handle any kind of waste. Still, the company wanted to ensure its customers were getting as much value from its robots as possible. 

Finding the best place for robots

“When implemented properly, AI and waste robotics can form a unique symbiotic relationship,” Duthilleul said. “However, these robotic arms can cost hundreds of thousands of dollars, so waste managers need to be sure they need a specific robot to extract more value from their materials before purchasing. Without gathering the data and analyzing it with AI ahead of robot implementation, MRFs lack visibility into waste flow composition.

The Robot Validator is a product that will help MRFs understand the impact robotics can have on their facilities. “These facility managers working with us and Waste Robotics are not locked into a specific robot or robot provider before they see the data they need to inform their decision,” Duthilleul said. “Instead, they can utilize an AI-led roadmap for analyzing the performance and cost of deploying different robotic systems to identify the best solution.” 

“[Greyparrot] does all the analytics and process control. Where it makes sense, we come in and deploy robots in situations within the recycling center where robots are needed,” Camirand said. “We have the data now because Greyparrot has been measuring and is able to say, ‘There are that many products going by. Maybe if you put a robot there and capture these products, it’d be making money.’”

Initial feedback on the system

Since rolling out this product, Waste Robotics has already seen customers use the Robot Validator and then decide to deploy full systems. 

“We’re really happy about this because it’s a very clever way to really reassure the client. Now, we’re not installing a robot and just guessing what the robot will do,” Camirand said. “We’re actually installing robots knowing full well, or with very high expectations, how the robot will perform. So we can guarantee performance to the client. They feel really reassured that we’re going in with the right dataset and that the performance will be met.” 

“Initial client feedback has been highly positive. In one Canadian materials recovery facility, the system identified significant recovery opportunities, detecting over 1,260 valuable objects per hour being lost to the residue line. Of these items, more than half were classified as highly valuable, and 90% were suitable for robotic recovery. This validation process helped the facility justify and implement robotic sorting solutions that measurably improved their material recovery rates,” Duthilleul said. 

Future plans for the partnership

Waste Robotics and Greyparrot see many opportunities to grow from each other as the companies offer complementary technology.

“We’re really in the first few innings of the partnership. But together with the Waste Robotics team, we’ll continue to apply the insights generated by AI waste analytics systems to help waste organizations find the most impactful use cases for robotics,” Duthilleul said. “Given we’re both rapidly growing, international companies, we have a very aligned mindset and the ambition to impact the waste management ecosystem.” 

“We can learn from each other,” Camirand said. “[Greyparrot] captures a lot of data, and that’s with detection AI. We do a lot of material handling through the robots, and that’s more motion AI, or gripper AI. These combined can become a real force.”