Automating pharmaceutical laboratory environments
Manual sterility tests are a routine and critical part in laboratories, yet they introduce contamination and inefficiencies owing to the high volume of daily tests. Automation has not yet taken over because robots lack the flexibility required to perform life-critical tasks. As a result, creating medical, pharmaceutical and healthcare products continue to rely on human operators, with the process being further slowed down by time-consuming quality checks and regulatory signoffs. The EU-funded TraceBot(opens in new window) project sought to further advance robotic operations in pharmaceutical laboratory settings, focusing on critical manual tasks such as membrane-based sterility testing. “These procedures are typically performed by skilled lab personnel under strict regulations and demand extreme precision as even minor errors can jeopardise safety or compromise results,” says project coordinator Maike Neumann.
Safe, explainable and certifiable robotic operations
TraceBot addresses key challenges in pharmaceutics by ensuring full traceability of robotic actions, allowing every step to be logged, reconstructed and verified. The project also enables robots to understand and reason about their actions in a way that humans can easily comprehend. Furthermore, future regulatory certification is supported by providing verifiable digital evidence of the robot’s decisions and performance. “While most laboratory robotic systems focus on automation and throughput, TraceBot prioritises cognitive capabilities. It is built to not only execute tasks but also to understand and explain the reasoning behind the tasks performed, going beyond mere action execution,” emphasises Neumann. What also sets TraceBot apart is its ability to justify decisions in real time – an area where most current systems fall short as they lack the capacity to explain or reason through their actions. “Designed with pharmaceutical environments in mind, TraceBot is built to meet the stringent demands of traceability and accountability,” adds Neumann.
Combining sensor, AI and digital twin technologies
In particular, the project consortium has developed an innovative robotic system that combines advanced sensors for precise perception, AI-driven action planning and error detection and a digital twin that monitors and evaluates robot actions in real time. “The digital twin goes beyond being a virtual replica of the robot; it acts as a cognitive monitoring system, synchronising in real time with the physical robot to reflect its actions and sensor feedback,” explains Neumann. By understanding the context of the tasks, it can detect inconsistencies or errors during execution and use symbolic AI for semantic reasoning to assess whether the robot’s behaviour aligns with expected procedures. Generating human-readable justifications for its decisions, it also supports after the event analysis and advances the development of explainable AI in safety-critical environments.
Broad industry impact
TraceBot brought together an interdisciplinary team of experts from robotics, AI, software verification, and lab automation, working alongside end users and industry regulators to ensure practical relevance. While initially focused on sterility testing, the project’s core technologies such as the cognitive digital twin, explainable robotics and traceability framework are applicable in other safety-critical domains, including compounded medicine, diagnostics and biotech manufacturing. Reflecting its innovation and industry impact, the TraceBot demonstrator has already earned recognition as it won the prestigious RAYA Award(opens in new window) for future software trends in pharma.
