Disassembly is an essential operation in many industrial activities including repair, remanufacturing and recycling. Disassembly tends to be manually carried out - it is labour intensive and usually inefficient. Disassembly requires high-level dexterity in manipulations and thereby can be more difficult to robotise in comparison to the tasks that have no physical contacts (e.g. computer visual inspection) or simple contacts (e.g. cutting, welding, pick-and-place). Robotic disassembly has the potential to improve the productivity of repair, remanufacturing, recycling, all of which have been recognised as key components of a more circular economy. The existing procedure and state-of-the-art techniques for disassembly automation usually require a comprehensive analysis of a disassembly task, correct design of sensing and compliance facilities, efficient task plans, and a reliable system integration. It is usually a complex, expensive and time-consuming process to implement a robotic disassembly system. This project will develop a self-learning mechanism to allow robots to learn disassembly tasks and the respective control strategies autonomously, by combining multidimensional sensing and machine learning techniques. This capability will help build a more plug-and-play disassembly automation system, and reduce the technical difficulties and the implementation costs of disassembly automation. It is expected the next generation industrial robotics can be adopted in more complex and uncertain tasks such as maintenance, cleaning, repair, remanufacturing and recycling, where many processes are contact-rich. Disassembly is a typical contact-rich task. The Principal Investigator envisages that self-learning robotic disassembly will provide key understandings and technologies that can be adopted to the automation of other types of contact-rich tasks in the future to encourage a wider adoption of robots in the UK industry.

Developing and improving our R&D and manufacturing capabilities to prepare greater numbers of higher quality crystalline materials has become a growing societal and hence industrial need. This requires higher levels of precision and speed throughout the R&D development cycle to meet the evolving needs for precision crystals in fine chemical's sector such as for pharmaceuticals, agrochemicals and additives. For example, a more differentiated product range is expected to be produced with a significantly faster molecule to patient journey, in much smaller volumes and at significantly lower costs. For pharmaceuticals, this will provide a wider range of more targeted medicines and dosage forms, ensuring the delivery of patient-targeted dosage forms with much improved safety and efficacy, hence enormously benefiting economy, environment and society. Such an increase in the multiplicity of crystalline products demands the implementation of digitally-enabled and AI technologies as highlighted in UK government policy and global initiatives. The surface properties of crystals are very important for the digital design and manufacture of precision particles via solution crystallisation. Control of the surfaces expressed on crystalline particles represents a critical objective for the fine chemical industry which manufactures ca. 70% of their ingredients in solid (crystalline) form. These crystals have their unique shapes and surface chemistry which, when variable, can impact adversely upon product quality and performance. Specifically, the effective digital design of such products and the associated processes for their manufacture demands a detailed knowledge of surface properties of the product's formulation ingredients. Currently there exists a critical gap to relate the measurable properties at the molecular and single crystal levels to the behaviour and performance of the same material when it is manufactured or used in particulate form. This perspective demands the development of a digitally-enabled platform which is able to characterise, monitor and control crystal size and shape. However, existing crystal shape descriptors available with current commercial particle measurement systems have limited capabilities and the corresponding algorithms tend, unrealistically, to be based upon the assumption that non-spherical crystals can be treated as spherical ones. Therefore, the development of advanced process-inspired analytical tools, particularly of AI-based approach, combining with first-principle, shape-based models are clearly needed. Such approaches are important in order to ensure that the UK's research-led fine chemical and pharmaceutical industry continues to provide outstanding international leadership in product development and manufacture so maintaining and enhancing its global competitiveness. The proposed research will apply machine learning based upon crystal morphology prediction (forward engineering) to map from 2D in-process microscopy data back to a description of a crystal's 3D shape (reverse engineering) and, through this, to its functional surface properties. This will enable the design and control of more efficient and agile manufacturing processes for crystalline fine chemicals, delivering precision crystals with a much tighter specification in terms of their size and shape than is currently feasible, hence resulting in products having more consistency, less variability, higher quality. The outcomes will be a digital platform of crystal shape characterisation and process dynamics control for precision particle manufacture. The approach developed will be shared through academic collaboration (such as the CMAC Hub, INFORM2020, Cambridge Crystallographic Data Centre, Imperial College etc.) and with industry (Infineum, Keyence, Pfizer, Roche, Syngenta etc.) and also extended in due course more widely, expecting potentially enormous economic and societal impact.