What can we expect from AI in automation and control?

The term artificial intelligence covers the science of intelligence generated by machines attempting to mimic the cognitive functions of humans. It is already widely used in our everyday lives and has enormous potential to affect us in the workplace, so what can manufacturing expect from AI in automation and control?

The term AI has been around since the 1950s and it covers a multitude of sciences from advanced analytics and machine learning to natural language processing and neural networks. For example, natural language processing gives us speech access to Amazon Alexa, Siri and Google Assistant and other devices.

AI in everyday life

Uses include banking for fraud prevention, event simulation and customer support and in finance for credit decisions,  risk management, and process automation. In healthcare, it analyses medical scans and screening, controls surgical robots, and predicts medication outcomes. In Internet marketing, it tracks where users go and gives them information from advertisers on products or sites they may like. It is also widely used in agriculture to help yield healthier crops, control pests, monitor soil and growing conditions.

Furthermore, it runs our smartphones and search engines and is behind facial and speech recognition, systems. An AI-driven smart Chatbot can help save time and human effort by automating customer support. Chatbots are computer programmes designed to converse with human users, and many people speak to telephone Chatbots without realising it.

Internet users send around 280Bn emails each day, and according to Statista, over 54% of all these are spam. That they did not all appear in our inboxes is a direct result of filtering using AI.  Web browsers use AI to search and index the Internet and identify fraudulent sites or offensive posts on social media.

According to management consultants McKinsey & Co. AI in automation and control is transforming businesses and contributing to economic growth through productivity improvements. It will also help address societal challenges in areas from health to climate change and transform the nature of work and the workplace. As a result, some occupations will decline, others will grow, and many more will change.

Benefiting from AI in automation

AI is interesting for industry and manufacturing as-is offers new ways to control machines and increase productivity. It also creates new business opportunities, as in the case of Rolls Royce. Their old service model was the simple repair and fixing or engines when they needed servicing.

Now, with thousands of jet engines are in use around the world, each year Rolls Royce receives more than 70 trillion data points from their in-service fleet. Using advanced data analytics, industrial AI, and machine learning enables them to unlock design, manufacturing, and operational efficiencies within Rolls-Royce. Moreover, by identifying unknown or unexpected occurrences they create new service propositions for customers around improved servicing intervals and fuel efficiencies.

However, most current manufacturing applications are on a smaller scale and centred around preventative maintenance, automated guided vehicle and virtual and augmented reality and machine vision.

Several factors have come together to raise the awareness and use of AI. AI depends on the ability to gather and process large volumes of data at high speed. Firstly, IoT is increasing the volumes of data users can access, and the cloud makes storage easier and cheaper. Next, increased computing power satisfies the demand for real-time information.

Advanced analytics algorithms enable computers to identify patterns, identify issues earlier and make better predictions. The processing of data can take place in the cloud, in-house or using local edge processing, or all three at once.

Finally, chip makers are implementing AI technologies on small-scale field-programmable gate arrays (FPGAs) and CPUs. This enables third-party designers to configure and develop tailored AI applications faster and at a lower cost.

Is it intelligence or machine learning?

AI in automation is the result of several AI sciences coming together. Of these, the most important are:

Advanced analytics (AA): within AI, advanced analytics provided by a wide range of high-level tools for delving deeper into available data. Working with big data to provide predictive analytics and use behaviour patterns, it helps users to understand why something happened and what is likely to happen again. One of the most important tools for industrial automation is the use of machine learning.

Machine learning (ML) involves computers discovering how they can perform tasks (better) without being explicitly programmed to do so. Computer algorithms analyse and learn from data received and improve from experience, without human intervention. With more data to analyse, the process will optimise over time. Popular examples of machine learning include autonomous vehicles, and speech and facial recognition systems.

Algorithms for machine learning are computer models formed from a finite sequence of instructions that define the process. In machine learning applications there are several standardised machine learning algorithms that are useful for solving most problems. There are many ‘standard’ algorithms available for users.

Deep learning  (DL) is part of a machine learning algorithm using artificial neural networks to learn by delayering large complex data sets.

Artificial neural networks (ANN) are computing learning algorithms designed to emulate human thinking.

Data mining is the automatic or semi-automatic process of identifying unknown patterns from large data sets.

Benefits or AI in control and automation

As with the Rolls Royce example, AI in automation allows users to understand their business in ways not possible before. For example, to:

– Enhance forecasting and decision making

– Innovate new products and services

– Improve how machines function and coexistence between machines and with humans

– Analyse and help users make sense of complex events

– Improve operational effectiveness and reduce downtime and increase production

– Continued improvement to machine processes including machine vision, preventative maintenance, virtual reality and augmented reality

Finally, it is important to realise the path ahead is not always straightforward, as is apparent from the difficulties of developing autonomous vehicles / self-driving cars. At the moment, AI lacks an element of common sense. The challenge for programmers is that common sense is neither rule-based nor entirely logical: it is something we grow up with. AI common sense computing will happen, it is just a matter of time.



Leave a Reply