Engineers at Swansea University have developed software that enables computers to learn from their mistakes.
X1 Recall, a self-learning, web-based software, uses Artificial Intelligence (AI) to help manufacturing companies eliminate waste and improve quality control in their production processes. The system is said to give computers the power of thought and reason, and to help them to learn from their own experience.
The system is the result of a research project undertaken by Dr Rajesh Ransing, a Senior Lecturer in the Universitys School of Engineering. The first phase of the research was funded by the Engineering and Physical Sciences Research Council (EPSRC), and the latter stages of the project have been supported by a 200,000 Knowledge Exploitation Fund grant from the Welsh Assembly Government with equal participation from a number of Welsh companies.
Dr Ransing said: Reducing the amount of scrap produced is one of the biggest challenges facing the manufacturing industry. Scrap wastes materials, energy and time, and the cost of reworking products can be a major drain on company resources for small companies as much as for large multinationals.
The ultimate result of our research will see the commercial development of self-learning software that enables industry to produce defect-free components right the first time, providing a quality control process that is by its very design continually improving.
X1 Recall has the potential to be used in any manufacturing situation where the production process relies on close management of a range of factors and variables, such as temperature, chemical composition, and so on. The software collates and analyses cause and response data from the manufacturing process, effectively looking at how changes to different factors affect the quality of the end product.
Where malfunctions, machinery breakdowns or faulty products result, the computer can examine the potential causes and recommend appropriate adjustments to the settings and variables. The recommendations are based on empirical evidence collected and learnt over time.
Dr Ransing said: What makes this research unique is the fact that it doesnt rely on experts to interpret the data and analyse the best course of corrective action. The software makes the computer think about the problem and reason why the issue has occurred. And all the time it is storing up information to help inform its future decision making.
The technology behind the system has been tested and proven in collaborative partnership with companies including Rolls Royce, the Cast Metals Federation (CMF), and several smaller businesses. Dr Ransing is now developing a prototype for use in industrial settings via a University spin out company MetaCause Solutions Ltd.
The UKs manufacturing industry faces stiff competition from other countries, particularly those with lower cost bases such as China and India. The opportunity to generate substantial cost savings by reducing waste and improving quality cannot be over-emphasised.
As part of the deal with the Knowledge Exploitation Fund, the Cast Metal Federations member companies who wish to use the software when it becomes commercially available will have a free licence granting access to the technology for five years at a small, one-off fee, in return for their participation in the commercial development process. This will give UK companies an edge over their international competitors.
After all, said Dr Ransing, the project has really been funded by the British tax payers through EPSRC and KEF, so it makes sense to ensure that the benefits stay in the UK. I am confident that X1 Recall has the potential to save millions if not billions of pounds for the British economy.
European companies can also use the technology at discounted rates if they join the programme, although this is subject to terms and conditions.
But Dr Ransing isnt focussing solely on industry. He sees the next stage of the research integrating X1 Recalls abilities with the NHS, facilitating more consistent diagnoses and treatments for patients, particularly in terms of recommending medical and pharmaceutical interventions.
Hospitals rely on pharmaceutical companies to advise on the appropriate dosage of their products in treating different conditions, but doctors very often develop their own gut feeling on treatments, perhaps increasing or decreasing dosages according to their experience of the drug.
However, if doctors move on to different roles, hospitals lose their experience and knowledge. X1 Recall will have the ability to analyse patient records across the NHS, learn from the combined experience of all doctors, and recommend appropriate courses of treatment based on the collated evidence of what has and hasnt worked in the past.
Essentially, what we hope to do is take all the available data from patient records symptoms, responses to treatments, etc and use the X1 Recall technology to enhance doctors decision making process," said Dr Ransing.
Evidence based suggestions on the precise course of treatment will also be possible, although the software will never replace doctors. No matter how intelligent computers become, no matter how much they can think for themselves, there will always be a need for the human touch in healthcare provision, he added.