5 questions about artificial intelligence and robotics technologies

Why is the rise of artificial intelligence (AI) technology so important in material recovery facility (MRF) operations?

Some of the biggest challenges of recycling sorting is that it is still a very costly, manual process. Most facilities are run chronically understaffed due to low unemployment. [With AI technology], we can sustainably replace very scarce labor, pick faster, more accurately, around the clock at a sustained rate with little operational upkeep. Lastly, with computer vision and machine learning, we are capturing and sharing data on material streams enabling recycling operations to gain transparency on the value of their lines, as well as measure productivity.

How does AI work on a practical level?

AI is the process of getting information and applying rules for using this information, applying logic, self-correction and learning to execute a task. The core problem that AMP has solved is teaching a machine to distinguish individual objects that may be dirty, smashed or torn in the midst of a very cluttered environment. That identification problem is quite difficult, as it means the machine must truly understand what makes a carton different than PET (polyethylene terephthalate), which is different than HDPE (high-density polyethylene), despite all of the variations that may be present. With artificial intelligence we solved this with the development of AMP Neuron™, our AI platform.

What are the main uses for AI and robotics in the recycling industry?

Using advanced computer vision and machine learning, the platform trains itself by processing millions of material images. It teaches itself to look for different visual attributes such as size, color and texture. It learns many of the same things a person would learn: PET is shiny, and aluminum has a gray circle on the bottom. Due to the learning-based approach, it learns from experience, not only getting better, but also identifying more specific categories of material. AMP Neuron can identify distinct materials with AI and discern what to do with each fraction, guiding robots within our AMP Cortex™ system to precisely sort consistently at a much faster rate and with greater accuracy than a person.

How does the incorporation of robotic technology affect manual sorting operations?

The modularity of our AMP Cortex™ robotic sorting system drops into the general footprint of a sorting station, making it incredibly scalable by adapting to existing operations. The installation is intended to be done within existing operations without costly retrofit and extended downtime, often over a weekend. We train on a given operation’s material stream in advance to jumpstart the integration, striving to be up and running as soon as possible. As far as working with or replacing manual sortation, it really depends on the customer operation and their goals. Our systems are designed and scalable to solve the labor challenge overall as a replacement. But due to the system’s modularity, robots can co-locate with manual sortation as well.

Where do you see AI technology headed?

We see a future (and it is happening fast) where facilities are increasingly automated at low cost, reducing operational friction and creating even more value. We also see the potential for new types of high-value commodity categories created as our AI is precise down to the consumer packaged good (CPG) brand level and even the UPC and SKU level. As our technology constantly adapts to new material types in the stream even more value will be created for the circular economy and its stakeholders.