From Autonomation to Autonomy

Since the introduction of ‘jidoka’ (自動化), industries have automated many processes through automation. But, they still do require human intervention to some extent and are programmed machines. Autonomy is the next step for the industry, and an important step in reaching that goal is better utilization of Artificial Intelligence (AI) and Robotics. Today, AI is usually considered a buzzword in many industries. It is reasonable to simply call is as a ‘buzzword’ to some extent, because at times it is applied in applications simply to boost the ‘marketability’ of things and it isn’t essential. But, tools like deep learning and machine learning when utilized fruitfully with appropriate data, have the power to drastically impact the performance, robustness, or even the general operation of any product in a positive manner. It is a challenging problem to explore, as full autonomy is hard to achieve, but there have been many advances in the past few years that are gradually building the rungs of the ladder to reach that goal.

In the past decade, AI, an umbrella term covering Machine Learning, Deep Learning, Data Analytics, etc. has seen tremendous growth and deployment in nearly all industries, manufacturing included. The advancement in Neural Networks has fueled the development of many new technologies in the past decade along with the growth in our computational capabilities. Manufacturing too has seen a lot of changes over the past decade with Industry 4.0 technologies, like robots, IoT, and Robotic Process Automation (RPA). The changes in both fields over the years have reduced the gap between them, making it easier to integrate these technologies. Especially, with the plethora of data available in manufacturing today, the possibilities become endless. 

Today, companies in a wide range of industries are trying to integrate analytics and data to improve their operations, as was seen in Rockwell Automation as well. Data Analytics can be used to understand yield data and get predictions and suggestions to improve processes, anomaly detection in time series data from factories and machines, etc. Analytics can also be used to attain ‘Heijunka’ by analyzing productivity, market demand and optimizing the manufacturing yield. Analytics techniques can enable better inferences and understanding of otherwise complex data, which can correspondingly help with leaner operations, reduced costs, and swift maintenance.

AI is usually compared to humans, and the limitations are obvious, but there are various tasks and areas in which AI tools can do a much better job. Vistra, a major U.S. power producer, had a problem. For its plants to operate efficiently, workers had to continuously monitor hundreds of different indicators, tracking temperatures, pressures, oxygen levels, and pump and fan speeds — and they had to make adjustments in real-time. The process involved a huge amount of complexity, and it was too much for even the most skilled operator to get right all the time. To address this challenge, the plant installed an AI-powered tool — a heart-rate optimizer — that analyzed hundreds of inputs and generated recommendations every 30 minutes. Result: a 1% increase in efficiency. The increment may not sound like much, but that itself could lead to millions of dollars in savings. This is just one example where AI tools perform much better than humans, owing to the complexity of the task. Advancements in Computer Vision due to deep learning allow for much better defect detection and performing quality checks on manufactured products, and even in between the processes to improve quality and the takt time.

Autonomous Robots and Manufacturing

Machines work tirelessly to mass-produce automobiles at the Fanuc headquarters just outside Tokyo, Japan (Image Source: Google)

Industrial robotic arms have long matured and become an integral part of the manufacturing industry, but most of them are still not autonomous. Autonomous systems, although separated by a thin line from automation, are capable of adjusting and optimizing performance on the fly and enable customization at the point of manufacture without human intervention. In the not-so-distant future, self-sustaining, fully autonomous factories with interconnected machinery that communicates, shares information and manages the complete production environment are possible. Ralf Hagen, an engineering manager at Nestlé Deutschland, says a key advantage of autonomous systems is that they handle security and maintenance behind the scenes.

"We are surely not going to have the number of skilled systems administrators in our company to do this because we are a production company, we are not an IT company. We are not an infrastructure company. We are just a simple producer of food. We want systems to be like a smartphone: Everyone has one in their pocket but doesn't have to think about how it works—it just does. On the industrial side, we can learn from this consumer experience." - Ralf Hagen

Many factories are already running pilots of such systems, and it might even take the transformation of the industry beyond industry 4.0. Autonomous assembly and manufacturing lines would essentially allow for continuous, high quality and large volume production, it would push aside any politics owing to unions, and increase the safety in industries where the risk of injuries is high for humans.

The Big (Realistic) Picture

Despite all the technologies available in today’s time and the limited areas already touched upon, a raft of opportunities is yet to be explored. Moreover, even though it may seem like AI is bound to make things successful, various factors make a company successful at using AI, such as the people, data availability, academic partnerships, governance, etc. There have been various studies that show the gains that can be made with employing AI and related technologies, it doesn’t guarantee success. Apart from these, it takes a lot of financial resources for the infrastructure to enable these technologies, and retrofitting in existing industries could increase the costs by a lot. Owing to this the industry leaders have the advantage here compared to others and the gap between the leaders and the rest could widen the gap making it a monopolistic market. 

AI and the Future of Manufacturing (Image Source: Google)

Looking at things from a different perspective, going from autonomation to automation to autonomy, although we may have increased the overall productivity and quality in many industries, at the same time one thing that is constantly reduced in the transition is human involvement. It is yet to be seen how big of concern the growing use of AI in an industry as fundamental as manufacturing is for the future, but there are boundless opportunities for development and increased earnings.

References:

1. “7 Ways Data Analytics Can Transform the Manufacturing Process .” 7 Ways Data Analytics Can Transform the Manufacturing Process (aimpointdigital.com)
2. “What Makes a Company Successful at Using AI?” https://hbr.org/2022/02/what-makes-a-company-successful-at-using-ai 
3. “Top 7 Deep Learning Applications in Manufacturing in 2022.” https://research.aimultiple.com/deep-learning-in-manufacturing/ 
4. “Moving from automated to autonomous manufacturing | HPE.” https://www.hpe.com/us/en/insights/articles/moving-from-automated-to-autonomous-manufacturing-2104.html
5. “Autonomous Manufacturing Creates Reactive Production - ASME.” https://www.asme.org/topics-resources/content/infographic-evolution-of-autonomous-manufacturing
6. “Toward smart production: Machine intelligence in business operations | McKinsey.” https://www.mckinsey.com/business-functions/operations/our-insights/toward-smart-production-machine-intelligence-in-business-operations
7. “17 Remarkable Use Cases of AI in Manufacturing | Birlasoft.” https://www.birlasoft.com/articles/17-use-cases-of-ai-in-manufacturing
8. “‘Lights-Out’ Manufacturing: Taking Uncertainty Out of the Equation - Ambyint.” https://www.ambyint.com/resource-item/lights-out-manufacturing-taking-uncertainty-out-of-the-equation/