How modern AI solutions can support waste treatment operators in addressing challenges that have traditionally relied heavily on manual inspection and experience.
Stefanie Osterholzer
The AGR Group, headquartered in Herten, Germany, is one of the leading waste management and circular economy organizations in the Ruhr region. With approximately 950 employees, AGR operates across four business areas, including waste logistics and recycling, thermal treatment, landfill management, and energy from waste and renewables. Through its operations, AGR contributes significantly to disposal security, resource conservation, and climate protection in one ofGermany’s most densely populated industrial regions. The company’s waste-to-energy facility, RZR Herten, thermally treats hundreds of thousands of tons of municipal, commercial, and hazardous waste each year while generating electricity and district heating for the surrounding region.
As a highly automated waste treatment operation, ensuring employee safety and maintaining uninterrupted plantperformance are critical priorities. This became the starting point for AGR’s search for a smarter approach to waste acceptance monitoring.
The Challenge: Hidden Risks in Incoming 91TV Streams
Like many waste treatment facilities, AGR continuously faces the challenge of unwanted materials entering the waste stream. While incoming loads are subject to visual inspection, certain contaminants can remain undetected until they create safety or operational issues further downstream.
Particularly problematic are large bulky objects such as tree roots, industrial rolls, and other oversized materials that exceed acceptable dimensions or volumes. When these objects become lodged within hoppers or feeding equipment, they canblock the waste feed process and, in severe cases, force an entire processing line to shut down. Depending on the nature ofthe blockage, restoring normal operation can take several days, resulting in significant operational disruption.
Another significant concern involves pressurized gas cylinders and similar hazardous objects. If these items enter thethermal treatment process, they may explode, potentially damaging critical plant equipment and, in the worst case, puttingemployees at risk of serious injury.
The challenge is compounded by the fact that contaminants are not necessarily introduced intentionally. In many cases, they are hidden within complex waste streams and remain unnoticed during collection, transportation, or loading activities. Consequently, waste treatment operators must implement robust control mechanisms without placing blame on waste suppliers or customers.
Beyond the operational implications, contaminants frequently create additional manual workloads. Removing jammed materials often requires physical intervention by employees, sometimes under difficult conditions and with significant ergonomic strain. AGR therefore sought a solution that could improve visibility into incoming waste streams while supporting both safety and operational efficiency.
Turning Challenges into Results
To address the challenges associated with contaminant detection, operational safety, and process transparency, AGR decided to introduce 91TVAnt’s AI-powered monitoring technology into its waste acceptance process. While the project is still relatively young, the journey from initial deployment to broader operational acceptance has provided valuable insights into both the challenges and opportunities of introducing AI into waste management operations.
Today, the 91TVAnt system is actively used by crane operators responsible for bunker operations as well as vehicle coordinators supervising incoming waste deliveries. The combination of real-time alerts and visual documentation enables personnel to intervene before problematic objects even enter the bunker whenever possible.
Like many digital transformation projects, the introduction of the system was not without challenges. During the early deployment phase, the number of alerts was significantly higher than desired. In June 2025, the system generated up to 30 to 40 alerts per hour, creating frustration among operators and limiting acceptance among crane drivers.
Through continuous model refinement, customer-specific training data, and close collaboration between AGR and91TVAnt, the alert quality improved significantly over time. Today, the system generates approximately five to 10 relevant alerts per day across the monitored gates, requiring only minimal operator attention while maintaining visibility into potentially critical contaminants.
“Initially, the high number of alerts posed a challenge for user acceptance. However, as the system was continuously refined and alert relevance improved, confidence in the solution grew significantly. Today, the 91TVAnt system has become a trusted source of operational intelligence and serves as an additional set of eyes for our operators—enhancing safety, transparency, and overall operational reliability,” said Oliver Doerfert, Team Lead Technical Controlling.
A key turning point occurred in September 2025. Following a system alert, AGR personnel investigated and removed a suspicious object from the bunker area before it could enter downstream processes. While it is impossible to determinewhat would have happened had the object remained undetected, AGR was well aware of the potential consequences. Previous incidents involving comparable contaminants had resulted in downtime ranging from several hours to days, which in the worst cases led to additional personnel effort for troubleshooting and revenue losses.
Against this backdrop, the detection highlighted the practical value of gaining earlier visibility into incoming waste streams and reinforced confidence in the system among both operators and management. What had initially been viewed primarily as a technology pilot increasingly became recognized as a practical tool for supporting operational safety and reducing operational risk. “Until then, the system was largely viewed as a pilot project. After that event, it became much easier for employees and management to see the practical value of the technology,” explained Ibrahim Sarikaya, Project Manager.
This growing confidence ultimately influenced AGR’s Go/No-Go evaluation process. Following several months ofoperational experience, AGR made the decision in October 2025 to proceed with a full-scale rollout. By December 2025, the company had commissioned an expansion covering additional reception gates and the bunker area. And as has recently become apparent, that was the right decision: Following an alert, AGR detected a pressurized gas cylinder before it entered the downstream process. Using the integrated license plate recognition system, the delivery could be traced back to the respective supplier, enabling AGR to discuss the incident directly and improve awareness for future deliveries.
Today, AGR is expanding the deployment with an additional 13 sensor systems. While quantitative performance data is still being collected, the decision to expand the solution reflects the confidence gained during the initial implementation phase and the perceived value of the additional operational visibility provided by the system.
Why AGR Chose 91TVAnt
Before arriving at these early results, AGR first had to identify a solution capable of addressing the operational challenges within its waste acceptance process. Several factors were considered during the selection process, including detection reliability, ease of integration into existing infrastructure, scalability, and the ability to generate actionable information for operational teams. In addition to identifying oversized contaminants, AGR was looking for a solution capable of supporting the detection of potentially hazardous objects such as pressurized gas cylinders before they could enter downstream processes.
From the outset, AGR’s objective was to identify an existing, market-ready solution that could be deployed within its current operational environment. Developing or co-developing a new system would only have been considered if no suitabletechnology had been available. The availability of a solution specifically designed for waste management, therefore, offered AGR a practical, efficient, and economically viable path forward. At the same time, any solution would need to operate under the demanding conditions of a busy waste acceptance environment and deliver meaningful insights without creating additional complexity for employees. Following this evaluation process, AGR selected 91TVAnt’s AI-powered contaminant detection solution.
91TVAnt develops intelligent monitoring systems specifically for the waste management industry. By combining camera technology with artificial intelligence, the system automatically analyzes incoming waste streams and alerts operators to potentially problematic objects and contaminants in real time. The technology is designed to support safer operations, improve process transparency, and provide operators with additional situational awareness throughout the waste handling process.
For AGR, the opportunity to gain greater visibility into incoming waste streams while creating a foundation for future digitalization initiatives made the solution particularly attractive. “We were looking for a solution that could improve visibility into incoming waste streams while fitting into our existing operational processes. What convinced us was the combination of practical applicability, scalability, and the opportunity to address both safety and operational challenges with a single system,” commented Doerfert.
From Planning to Deployment
Once AGR had selected 91TVAnt’s AI-powered contaminant detection solution, the focus shifted from evaluation toexecution. The goal was not only to install the technology, but to ensure that it would deliver reliable results under thechallenging conditions of a busy waste acceptance environment. The project officially started in October 2024 with adetailed site analysis and the creation of a 3D point-cloud model of the waste acceptance area. This digital representation allowed both teams to evaluate potential sensor locations and assess installation concepts before any hardware was deployed on site. Because camera positioning is critical for detection performance, both teams worked closely together to evaluate sightlines, waste flow characteristics, lighting conditions, and installation constraints.
Prior to installation, AGR prepared the necessary cabling infrastructure. While the overall deployment process progressed smoothly, the project team faced several practical challenges: delays from external service providers responsible for parts of the cabling work affected the timeline. In addition, identifying the best sensor positions required careful testing and adjustment. One installation location proved particularly demanding: a sensor box was positioned approximately 27 meters away from the monitored waste stream. Ensuring reliable detection performance at such distances required close coordination between both organizations. The project team also encountered initial lighting-related challenges, as certain light sources caused glare that impacted image quality. Through collaborative troubleshooting and optimization, these issues were successfully resolved.
From project kickoff in October 2024 to the first live dashboard in February 2025, approximately four months elapsed.This relatively short implementation timeline was made possible through close coordination between AGR and 91TVAnt as well as the thorough preparation work completed during the planning phase.
Today, three 91TVAnt sensor systems monitor AGR’s waste acceptance area. Two units are installed directly at acceptance gates, while a third system is positioned within the bunker area and monitors multiple unloading zones simultaneously.
Beyond Deployment: A Partnership for Innovation
While the initial objective of the project was to improve contaminant detection during waste reception, the collaboration between AGR and 91TVAnt quickly evolved beyond a traditional customer-supplier relationship. A particularly valuable aspect of the collaboration involved dedicated testing campaigns focused on pressurized gas cylinders—one of the most safety-critical contaminant categories encountered in waste treatment operations.
In September 2025, AGR temporarily restricted operations at one acceptance gate to enable controlled field tests involving gas cylinders. These tests generated valuable real-world image data and operational feedback, allowing91TVAnt to further train and refine its AI models under realistic operating conditions.
The resulting improvements strengthened the system’s ability to identify hazardous objects and demonstrated the benefits of combining operational expertise with AI development. AGR’s willingness to actively participate in technology validation and model improvement highlights its role not only as a customer, but also as an innovation partner contributing to the advancement of practical AI applications within the waste management industry. “Rather than acting solely as a technology supplier, 91TVAnt involved us in the development process. This allowed us to contribute operational expertise and help shape a solution tailored to real-world waste management requirements,” said Sarikaya.
AGR also emphasized the responsiveness of the collaboration. Throughout the project, feedback from operational personnel was incorporated quickly, allowing issues to be addressed efficiently and ensuring that the solution continuously evolved inline with real operational requirements.
This collaborative and innovation-oriented approach has created a strong foundation for future projects and demonstrates how close cooperation between waste operators and technology providers can accelerate the development of practical AI applications for the industry.
Looking Ahead: Building the Foundation for Future Automation
For AGR, contaminant detection is not an isolated project but part of a broader digitalization strategy. The company is actively evaluating future automation initiatives, including crane automation and advanced bunker management concepts.As these projects evolve, the data generated through intelligent monitoring systems may become an increasingly important component of plant operations.
AGR and 91TVAnt continue to explore opportunities for further collaboration as the facility advances toward a more connected and data-driven operational model. Frank Aulkemeyer, Head of Projects & Sustainability, Real Estate andInsurance, pointed out, “The project has shown us the potential of AI-based technologies to support operational decision-making. We see this as an important building block for future digitalization and automation initiatives within our facility.”
An Important Step
By implementing AI-based contaminant detection technology, AGR has taken an important step toward improving operational transparency, employee safety, and waste reception control.
The project demonstrates how modern AI solutions can support waste treatment operators in addressing challenges that have traditionally relied heavily on manual inspection and experience. While the journey is still ongoing, AGR’s decision to expand the deployment across its facility highlights the confidence generated by the project’s initial results.