Supports real-time central command, remote firmware updates, and consistent performance across multiple sites through cloud-based infrastructure
Enables scaling to 1,000+ sites with Google Cloud AI Infrastructure and Compute Engine
Structures high-volume waste data for downstream use in compliance reporting and consumer insights with Cloud Data Fusion
Addresses current manual recovery limitations and improves material recovery by 10%, with potential future gains of up to 50%
TeknTrash, powered by Google Cloud, transforms waste sorting with its Robots-as-a-Service (RaaS) model that offers 24/7, lower-cost operations, boosts recovery accuracy by up to 50%, and provides vital data for compliance and sustainability.
Each year across the world, millions of tons of waste are sorted by hand in dirty and hazardous conditions, with waste and recycling jobs among the top five most dangerous occupations in the United States. At the same time, outdated recycling systems have become a major bottleneck to achieving a circular economy, just as Extended Producer Responsibility (EPR) laws are requiring businesses to account for their products from cradle to grave.
“If you go to a mixed recycling facility, you’ll see long picking lines where people are sorting different waste streams. It’s a very hands-on, dirty, and dangerous job, where injury rates are high,” says Michael Cox, commercial director of TeknTrash Robotics. “What’s more, next to no data is collected about the waste that comes in and what gets extracted for recycling.” TeknTrash aims to tackle those issues through a two-pronged approach: It automates the most dangerous and repetitive aspects of waste sorting that humans don’t want to do, helping keep workers safe. Meanwhile, it uses advanced imaging and AI to generate granular waste data and provide detailed insights for compliance, operational optimization, and sustainability reporting.
Google is doing so much more for us than just offering cloud space. Since the beginning, the Google Cloud team has been incredibly supportive of our mission.
Al Costa
CEO, TeknTrash Robotics
To do this, the company is deploying ALPHA (Automated Litter Processing Humanoid Assistant), a cloud-enabled humanoid robot designed to safely identify, sort, and record waste items in real time. To build the high-performance infrastructure needed for training AI models, processing hyperspectral imaging data, and remotely managing robots across hundreds of recycling facilities, TeknTrash turned to Google Cloud. "Google is doing so much more for us than just offering cloud space," says Al Costa, CEO at TeknTrash Robotics. "Since the beginning, the Google Cloud team has been incredibly supportive of our mission."
ALPHA uses hyperspectral imaging to scan and identify waste items, even when they are dirty or incomplete. Once identified, the robot can physically pick up and place each item into the appropriate recycling stream, while logging data about the material and product type, providing a level of visibility that would be impossible to achieve manually. The robots are guided by cloud-based command architecture: Rather than having all processors run on individual robots, ALPHA is cloud-connected, with heavy computation happening offboard. Compute Engine provides the high-performance GPU infrastructure required to process hyperspectral imaging data, run real-time classification models, and execute remote inference across multiple sites. "The power of ALPHA lies in its ability to learn and adapt, which is only possible because it's cloud-based," says Costa. "With Google Cloud, we can process vast amounts of imaging data offboard and remotely train ALPHA to recognize local recycling standards and product variations. That level of adaptability wouldn't be feasible without a robust cloud infrastructure."
The power of ALPHA lies in its ability to learn and adapt, only possible because it's cloud-based. With Google Cloud, we can process vast amounts of imaging data offboard and remotely train ALPHA to recognize local recycling standards and product variations. That wouldn't be feasible without robust cloud infrastructure.
Al Costa
CEO, TeknTrash Robotics
This setup is the foundation for the company's Robotics-as-a-Service (RaaS) model. TeknTrash's solutions are deployed and managed as a subscription-based service, with recycling plants paying for ongoing operations, updates, and data access. This model depends on remote deployment and real-time service delivery alongside operational intelligence derived from the continuous capture and analysis of high-volume data streams.
ALPHA captures over 6 megabytes of imaging data every second. We rely on Google Cloud to centralize and manage the full data stream from all our systems, from hyperspectral cameras to robot logs.
Al Costa
CEO, TeknTrash Robotics
ALPHA robots and their supporting systems generate considerable amounts of data from several different sources, including hyperspectral vision data, robotic movement logs, and conveyor belt camera streams. This all adds up to a total of over 49 terabytes of structured and unstructured data centralized and stored using Google Cloud Storage. "ALPHA captures over 6 megabytes of imaging data every second. We rely on Google Cloud to centralize and manage the full data stream from all our systems, from hyperspectral cameras to robot logs. Without that infrastructure, the entire safety and intelligence layer would break down," Costa explains.
TeknTrash uses Cloud Data Fusion to structure, clean, and integrate the vast amounts of recycling data generated by its solutions. "Right now, there's almost no data collected about the waste that comes in and what gets extracted for recycling," Cox explains. "ALPHA will enable tracking of all those individual items, giving companies the granular insights they need for Extended Producer Responsibility and cradle-to-grave reporting. That level of detail hasn't existed in the industry until now."
As waste items are identified and sorted, detailed metadata is logged and sent to Google Cloud. Cloud Data Fusion orchestrates this incoming data from multiple sources, such as hyperspectral cameras, headset training sessions, and robot logs, and transforms it into structured formats.
The data can then be used for internal performance analysis, regulatory reporting such as Extended Producer Responsibility (EPR) compliance, and commercial insights for FMCG companies. Using ALPHA robots to collect real-time data for reporting saves companies hours of manual work, increasing efficiency, while giving them powerful insights they've never had before. "What we're doing is turning raw waste streams into environmental and market intelligence," says Costa. "We're not just reporting on materials, we're also providing data on product use."
TeknTrash currently has three active deployments: VR headsets, hyperspectral cameras, and conveyor belt monitoring systems, with all three solutions feeding data to the cloud for training, analysis, and operational benchmarking. A full deployment across multiple sites run by London-based waste management company Sharp Group is planned by the end of this year. This will serve both as a proof of concept and as a foundation for future rollouts across up to 1,000 recycling plants in the United Kingdom and Europe.
"Scalability is built into our modus operandi, and that's why our partnership with Google Cloud is so instrumental," Costa explains. "We need powerful servers — and that's where Google Cloud delivers. Deploying or deleting a server and starting fresh is really easy. With a thousand robots across just as many plants, without it we would be spending all our time setting up infrastructure."
We need powerful servers — and that's where Google Cloud delivers. Deploying or deleting a server and starting fresh is really easy. With a thousand robots across just as many plants, without it we would be spending all our time setting up infrastructure.
Al Costa
CEO, TeknTrash Robotics
The TeknTrash team is currently projecting a 10% improvement in recovery efficiency, with the potential to reach a 95% recovery rate in the future. "A human can sort around 30 to 40 pieces per minute. But people get tired, and that number drops to zero in just a few hours. Robots can work three shifts a day, enabling 24/7 operations," Costa says. The company intends to move beyond off-the-shelf robotics and develop a fully custom humanoid robot, purpose-built for the challenges of waste management environments. For this, they're exploring the potential of Gemini Robotics.
"Our five to 10-year vision is a robot that can handle every task in a recycling facility. It's an exciting time for robotics, and we're looking forward to seeing where the future will take us," Costa concludes.
Based in the United Kingdom, TeknTrash Robotics is a research company seeking to develop cutting-edge solutions to make the waste sorting industry safer and more sustainable.
Industry: Technology
Location: United Kingdom
Products: Google Cloud, Cloud Data Fusion, Cloud Storage, Compute Engine, Gemini Robotics