Inspection of parenteral products is essential in pharmaceutical manufacturing. Every container leaving a facility must meet strict quality standards to ensure patient safety and maintain efficient production.
Discussions about visual inspection often focus on hardware—cameras, lighting, handling systems, and increasingly sophisticated robotics. However, the true intelligence of a modern inspection system lies in its software, where much of the development effort and value are concentrated.
At its core, automated inspection follows a structured workflow:-
Stage 1: Container Handling and Positioning
The process begins with container handling. Before inspection, products must be positioned correctly. Automated systems manage containers differently from human inspectors, who naturally manipulate them to maximise defect visibility. Modern robotic systems increasingly replicate these movements, bringing key benefits of manual inspection into automated environments.
Stage 2: Image Acquisition
Next comes image acquisition. Human inspectors build visual understanding by examining containers against white and black backgrounds. Automated systems achieve the same objective through carefully designed lighting and camera configurations, capturing images from multiple angles and under different conditions.
Stage 3: Software-Based Image Analysis
Once images are captured, software takes over. Like the human brain, inspection software processes visual information through a combination of sequential and parallel operations. Each stage focuses on specific characteristics or defect types.
For example, an algorithm designed to detect glass particles evaluates features such as size, shape, movement, and contrast to determine whether contamination is present. Similar approaches are used to identify fibres, rubber particles, metal contamination, hair, cosmetic defects, cracks, scratches, flange defects, and crimp cap damage. Because each defect presents unique challenges, effective detection requires a dedicated algorithmic approach.
This process is more comparable to human inspection than many realise. Some operators can reliably detect particles as small as 50 microns, while others consistently
identify only those above 100 microns. Performance varies based on experience, visual acuity, and training. The same principle applies to algorithms.
Stage 4: Continuous Learning and AI Enhancement
Algorithms can evolve and improve over time. This is where artificial intelligence adds value. For some defect types, conventional algorithms already perform exceptionally well, leaving limited room for improvement. For others, such as crack detection, machine learning can significantly enhance results. In our approach, AI functions as a learning layer within the inspection workflow.
Initially, it operates independently of inspection decisions, analysing data, identifying optimisation opportunities, and supporting the development of future algorithm generations.
Within our software environment, workflows are fully visualised. Users can review each inspection step, understand algorithm interactions, and monitor outputs throughout the process. AI modules appear as learning branches that collect data and generate insights without influencing production decisions until fully validated.
This distinction is particularly important in pharmaceutical manufacturing, where new algorithms must be rigorously tested, documented, validated, and approved before deployment in regulated environments.
Stage 5: Decision Optimisation and Waste Reduction
The ultimate objective is straightforward: reduce false rejects without compromising product quality.
Why does this matter?
Pharmaceutical manufacturing is fundamentally a value-creation process. Raw materials enter at one end, finished products leave at the other, and inefficiencies reduce overall value.
False rejects are a major source of waste in parenteral manufacturing. Every incorrectly rejected product consumes materials, energy, labour, and production capacity without delivering value to patients.
The impact can be significant!
By improving inspection performance through advanced algorithms, optimised software workflows, and carefully implemented AI, manufacturers can substantially reduce these losses.
The benefits extend beyond efficiency.
Reducing waste strengthens competitiveness, supports sustainability objectives, improves resource utilisation, and contributes to long-term industry growth.
Technology is not the end goal.
The real objective is to understand the industry’s importance and apply innovation to create safer products, more efficient manufacturing processes, and a sustainable future for pharmaceutical production.
ROVIS Robotic Visual Inspection System
By combining advanced inspection software with robotic handling systems that emulate human manipulation techniques, ROVIS brings together the strengths of manual and automated inspection.
The platform integrates container handling, image acquisition, defect detection, workflow visualisation, and AI-driven learning in a single environment. This provides manufacturers with greater visibility into inspection performance while continuously improving detection capabilities through data-driven optimisation.
Whether used for particle inspection, cosmetic defect detection, container closure integrity assessment, or other visual inspection applications, ROVIS supports higher quality standards, fewer false rejects, and more efficient manufacturing operations.
QPS Engineering combines extensive pharmaceutical expertise with capabilities in defect catalogue development, defect test set creation, AVI training, and inspection system validation. By helping manufacturers define and standardise defect libraries, we enable more consistent and reliable inspection across manual and automated environments.
Our experience developing and applying representative defect test sets provides valuable insight into defect behaviour, variability, and detection challenges. This expertise enables us to deliver solutions that are technically robust, operationally effective, and aligned with regulatory expectations. Operating at 12 units per minute, ROVIS is designed for small-batch
production and is particularly suited to re-inspection of AVI rejects, AQL sampling, and inspection of high-value ATMPs. ROVIS combines advanced inspection technology with practical pharmaceutical expertise to help manufacturers improve quality, reduce waste, and increase confidence in inspection outcomes. By integrating robotics, workflow-driven software, and AI-enabled learning within a single validated platform, it provides a scalable path to more efficient and reliable visual inspection. If you would like to learn how ROVIS can enhance your inspection processes, support your digital transformation strategy, or evaluate the role of AI within your inspection programme, contact QPS Engineering. Our team is ready to discuss your challenges and identify the most effective path forward.
About QPS Engineering
QPS Engineering AG, located close to Basel in Switzerland, provides advanced, GxP-compliant engineering and digital solutions for the Life Sciences, supporting pharmaceutical and biotech clients from concept to validated operation. Its four divisions reflect this expertise: Qengineering delivers project and process engineering with a strong focus on on-site validation and qualification activities through to PQ, aligned with its role in providing comprehensive technical consulting, project management, and qualification support; Qlabs develops specialized visual-inspection test sets for MVI, SAVI, and AVI; Qmeasure offers high-quality measurement and certification-relevant services; and QxTec provides AI-driven robotics, software, and visual-inspection systems ensuring product and process quality.
