Future Manufacturing: Young Professionals Research Showcase

Programme

Abdullah All Mamun Anik, AFHEA, Engineer and Doctoral Researcher, School of Computing and Engineering, University of Huddersfield

‘Capability modelling of robotic machining systems through a digital twin’

Research Synopsis: In the era of technological advancements, digitalisation, advanced manufacturing, and industrial revolution, the world is moving forward with intelligent manufacturing processes. Robotic machining is becoming very popular nowadays for their manufacturing capabilities. Manufacturing Engineers are making their production planning through manual check before robotic machining without any organised capability modelling. Traditional capability evaluation method is manual, and human centric. It is non-standardised and time consuming. There is a high risk of human error. Capability modelling of robotic machining system through bridges the gap between theoretical robot specifications and practical machining applications, driving innovation and competitiveness in automated and cloud manufacturing. In cloud manufacturing, Digital Twin are now established technology on the market and are commonly deployed for robotic machining. Capability modelling is a part of digital twin for robotic machining production planning, feasibility analysis, and demand estimation. However, digital twin technology does not have any capability modelling architecture with it for ensuring production plan and forecasting robot’s operational capacity. Existing research focus on bi-directional real-time data exchange without much effort on capability modelling. Creating a capability model for Digital Twin of robotic machining will enable autonomous cloud manufacturing. In this research, a capability profile of robotic system has been proposed and implemented. An industrially inspired component was used as a case study. This research work demonstrates data modelling, capability modelling, CAD file geometric data extraction system to build an intricate model for digital twin. The main components incorporate the judgement of the component design on the basis of size, machining envelope, operational details, cutting tools, cutting strategy, machining coordinate position, operational time, and data analysis. This paradigm extends transformative mechanism towards intelligent, data-driven manufacturing ecosystems for robotic machining.

Dr Chaoyue Niu, Research Associate, School of Computer Science, The University of Sheffield

In-Process Technology for Online Quality Measurement of Robotic Machining in High-Precision Manufacturing

Summary: Achieving high-precision robotic machining typically relies on costly position sensing (laser tracker) and post-process inspections (coordinate measuring machine - CMM). This research presents a data-driven framework that enables "Right-First-Time Assurance" of machining through in-process quality prediction. By fusing low-cost external force and vibration sensors and internal robot controller signals using advanced machine learning models (Mixture-of-Experts), the approach accurately predicts dimensional and geometric errors during machining. Tested across various tool trajectories and workpiece geometries, the models achieved a sevenfold reduction in mean absolute error, from 71.4 to 9.2 microns, compared with direct laser tracker measurements against CMM, with sub-5 ms/inference latency on CPU hardware. Regarding its relevance to the future of manufacturing, proving that low-cost sensor fusion can match high-end metrology provides the foundation for closed-loop parameter compensation and "Inspection by Exception." This shifts production toward scalable, zero-defect autonomous manufacturing, drastically reducing inspection overhead and re-machining costs.