Automated Digital Inspection for Asset Lifecycle Certification

Quality assurance during manufacture, combined with asset monitoring through service life, can help extract maximum useful life (whilst maintaining safety margins) from major engineering components used, for example, in aircraft, seagoing vessels and renewable energy structures. The role of NDT (non-destructive testing) is central to such quality assurance. NDT is now routinely built into the digital-twin approaches to lifecycle engineering employed in industry 4.0 manufacturing.
Our initial collaboration with TWI employed fixed robots to deliver NDT measurements, typically following pre-programmed paths, thus making them less suitable to handle and inspect parts with a significant geometrical tolerance or variability. We then moved to look at metrology and machine vision-based path correction, employing a novel real-time approach for industrial robots for sensor-based path correction. This approach has been very successful for manufacturing inspection of components up to several metres in scale. Beyond this, when we consider large components such as lifeboat hulls, aircraft wings, or wind turbine blades (scale of 10 of metres), and when we consider in-service inspection, there is a shift to using mobile robotic platforms to deliver the NDT measurements. Unfortunately, such mobile robotics technology is still in its infancy despite strong interest in autonomous systems (driven by interest in self-driving vehicles). At present, there are significant challenges for combining the kinematics between mobile robot platform base units with collaborative robotic arms allowing for safe working in crowded and often dynamic environments encountered in manufacturing and periodic inspection and repair operations. The challenges involve on-the-fly path planning from variable geometry surfaces, collision and obstacle avoidance, control of the NDT measurement process and the ability to adaptively respond to changing circumstances such that high-quality NDT measurements referenced to standard calibration procedures can be performed.
The aims of this project will be:
(1) to develop novel robot base manoeuvrability and control that is programmatically compatible with the control of industry-standard collaborative robot arms
(2) to investigate the role of AI-based processing for scene recognition to enable efficient NDT path creation on unknown geometries whilst maintaining collision-free operation in cluttered environments
(3) to use on-line machine learning-based data interpretation for the NDT measurements, thus allowing for in-process compensation/ remeasurement
The methodology will be a combination of simulation and practical experimental working. Simulation will be used to inform both robot path planning from the kinematics of the hardware, and also to understand the optimal NDT strategy dependent on sample, material properties and local geometry. The successful student will initially be trained in the latest automation and NDT capabilities in the new £29M SEARCH laboratory based in EEE. Working with the extensive and established team in SEARCH, the student will spend year 1 grounding in fundamental principles and completing a literature review and background state-of-the-art study. The students will then transition to bespoke facilities at TWI (Port Talbot) to continue their studies and develop the new hardware. Several industrial case study inspections will be conducted during this period, drawing from aerospace, naval and renewable energy application sectors.