Key points are not available for this paper at this time.
Abstract The adoption of robotics and automated solutions in life sciences R&D has accelerated in recent years, driven by the need to process increasing sample volumes, protect laboratory staff from hazardous substances, and manage financial pressures. Various automation systems, each with distinct levels of sample processing, transportation tasks, and data management, are available to meet specific application requirements, with liquid handling robots taking pivotal positions in these systems. However, current liquid handling robots, such as the Opentrons OT-2, lack integrated vision-based quality control, which limits their accuracy and reliability. This study presents an AI-driven computer vision model designed to enhance quality control in laboratory automation. By integrating the YOLOv8 object detection model with the OT-2, our model enables precise detection of pipette tips and liquid volumes, providing real-time feedback on errors, such as missing tips and incorrect liquid levels. Our results demonstrate the model's effectiveness and accessibility, presenting an affordable solution for improving automation in academic and research laboratories. This closed-loop system transforms the OT-2 into a robust tool for automated laboratory tasks, making it an accessible and cost-effective approach for enhancing quality control in laboratory automation and addressing a critical gap in available tools for resource-limited settings.
Khan et al. (2025) studied this question.