Will AI Take Over Production Processes?

By Birds.ai

Will AI take over production processes? According to a survey conducted by Forbes Advisor, up to 53% of businesses employ artificial intelligence to enhance their manufacturing processes, with 30% choosing to use it in their supply chains. 

Figure 1: Top Ways Business Owners Use Artificial Intelligence (Forbes Advisor, 2023)

So, yes! Whether we like it or not, artificial intelligence will soon have a big impact on supply chain operations and production processes. 

As Birds.ai, we witness an increasing trend of production companies introducing AI into their production with a main focus of improving the efficiency, safety and gathering more accurate data. Especially in high-volume production, we noticed that AI can make an enormous impact since the repetition is high and a small improvement in efficiency can have a big financial impact. 

Also the safety measures that need to be put in place for people working alongside automated processes are becoming more and more important and AI can help in decreasing the risks. 

What do AI models do?

The role of AI can be brought back to 3 main characteristics, being: Classification, Detection, and Tracking. On the basis of these 3 things corrective actions can be taken and/or predictive actions in some cases.

AI’s role for classification: 

Artificial intelligence can classify things by being trained by detecting how a user interacted with the content and elaborate into their own understanding. For example, supervised learning; the most common technique for machine learning. It’s a type of algorithm that is trained on a dataset in which each example in the dataset. For instance, consider a dataset used to predict whether an animal is a cow or not, each animal would be an example with features as input data and a label (cow or not) associated with it. 

One of the things that AI developers do is that they retrieve, preprocess and analyze huge amounts of data to train AI models. It can be simply explained by how humans do to learn new things faster, looking at examples; however, with machines, they require thousands of examples. 

 A Norwegian seafood company, Lerøy Seafood Group, decided to install cameras combined with their hyperspectral imaging arm from the research company Norsk Elektro Optikk into their conveyor belt to measure the amount of residual blood in the muscle. Fish that appear with a pink or red appearance because of stressing that got while being caught is believed to be lower quality. The operator will see information on each fish that was obtained by cameras on the monitor board's screen. The fish will subsequently be divided into various groups based on its color. For instance, the ones lacking a pink or red color will be taken off the conveyor belt and used elsewhere right away. 

AI’s contribution to detection in manufacturing:

Detection involves identifying the presence or location of certain objects or anomaly detection of the input; for example, images or videos. It answered the question of whether an object or feature is present. Oftens, operators use AI detection in production environments for quality control, security and surveillance systems. 

According to a research conducted by Mordor Intelligence, The AI Image Recognition market size was expected to reach 4.39 billion USD in 2023, and increase 2,48 times after 5 years into 10.91 billion USD. In reality, Statista provided the number for the market size in the Image Recognition itself which already reached 10.53 billion USD in 2023.

Foxconn, a company known for producing electronic devices for major brands like Apple, Sony, Nokia, and Nintendo, has incorporated Google Cloud's Visual Inspection AI into its factories since 2021. This machine learning program, introduced by Google, has greatly assisted manufacturers in identifying product defects, leading to a reduction in quality assurance expenses.

Let's delve into the smartphone production process as an example. Prior to the manufacturing phase, operators had to ensure the highest quality for a critical component, the Printed Circuit Board (PCB). Often, PCBs could develop issues like missing screws or solder bridging. Consequently, this new AI inspection system has succeeded in decreasing the defect escape rate by up to 10%, all while reducing inspection time for other components to just 0.3 seconds.

Another example on how AI-powered inspection for classifying is Birds.ai ‘s manufacturing line inspection projects. We trained the AI model to employ a counting method for detecting and tracking glass bottles as they move through an inspection line. With the improved performance, the client was able to optimize their production process with greater accuracy, safety and a reduction of waste.

AI’s Tracking:

AI tracking in the manufacturing process entails the application of AI to monitor and regulate numerous production-line elements for efficient operations, quality control, and quantity control. This is where Industry 4.0 comes into play, optimizing the entire manufacturing ecosystem through the integration of AI and data analytics

Why do we need tracking, you may ask? Because, sometimes we don’t just track for one or two objects, but hundreds of thousands of identical objects. Finding them won’t be easy like your typical Shell Game. 

Let's use the project from Birds.ai as an example. One significant challenge in mass production companies is that each section or step must be executed flawlessly due to the physical separation between these sections and the potential cost implications of minor defects or missing items. 

The issue of bottles falling or breaking due to jams is particularly relevant in this case. In certain areas, bottles may no longer be conveyed in perfect alignment, and to the naked eye, it becomes impossible to identify the location of the jam or determine how many bottles have been affected. By implementing the Birds.ai inspection system, operators can now ensure that the number of bottles entering a section matches the number of bottles exiting it, which made it easier for them to take a proactive approach to fixing these issues.

The predictive power of AI in Supply Chains: 

In addition, AI can be trained to predict and support businesses in making better business-decisions in their supply chain management, in terms of risk management, inventory management, demanding and forecast. Business forecasting is a combination of collecting past data and identifying current trends. There are two kinds of business forecasting models: Qualitative models and Quantitative models. 

Qualitative models are built strongly for expert-driven decision-making. The activities included doing market research and compiling enough thoughts of field experts on their general opinions (Delphi method). On the other hand, Quantitative models tried to remove the most human element from the analysis as much as possible such as the indicator approach, econometric modeling, time series methods.  With the rise of AI-powered inspection, more quantitative models can be developed and improved. 

Danone is one of the businesses that chose to use artificial intelligence into predicting the influence of their marketing campaigns to the demand of their products, which traditionally would be involved by many external influences such as market dynamics, consumer behavior, seasonality…etc. The new technology, however, allows Danone to apply machine learning capabilities to analyze past data pertaining to their marketing initiatives and demand trends, which resulted in an increase of 6% on their ROI (Return on Investment) in the first year, 30% reduction in lost sales.

In 2021, the AI development team of Birds.ai started a project in their Horticultural section called “Strawberry AI” purposeful for farmers and managers for their strawberry farms.

The project goal is to develop an AI model into predicting the harvesting time to avoid wasted early and late strawberries arrival. The algorithm trained in the previous step is capable of generating valuable predictions regarding maturity, shelf life, and suitability within the supply chain. 

In general, with its predictive capabilities transforming decision-making, it’s clear that AI is a competitive advantage. Companies should take into account the benefits of this new technology to stay current with trend and develop a more robust supply chain management system. It empowers manufacturers to streamline their operations, strengthen the product quality and optimize resources. 

The future of production processes is increasingly intertwined with artificial intelligence, and businesses that embrace it sooner are likely to thrive in the race of the new era of manufacturing. As for us Birds.ai, we see ourselves fully engaged in this future and focussing on how to help companies make better and more sustainable decisions. 

References

adiyanthy. (2019, March 9). ';;'. ';;' - YouTube. Retrieved October 22, 2023, from

https://medium.com/@adiyanthy/basics-of-machine-learning-2774c222df67

AltexSoft. (2022, July 23). AI in Manufacturing: 5 Use Cases. AltexSoft. Retrieved October 12, 2023, from https://www.altexsoft.com/blog/ai-manufacturing/

Haan, K. (2023, April 24). How Businesses Are Using Artificial Intelligence In 2023. Forbes. Retrieved September 25, 2023, from https://www.forbes.com/advisor/business/software/ai-in-business/

Manufacturing Lines Inspection — Birds.ai. (2023). Birds.ai. Retrieved September 25, 2023, from https://birds.ai/manufacturing

Norwegian fish filleting gains hyperspectral quality measure. (n.d.). Imaging and Machine Vision Europe. https://www.imveurope.com/news/norwegian-fish-filleting-gains-hyperspectral-quality-measure

University of Colorado Boulder. (2022, October 2). . . - YouTube. Retrieved September 28, 2023, from https://www.google.com/url?sa=t&rct=j&q=&esrc=s&source=web&cd=&cad=rja&uact=8&ved=2ahUKEwjjqsyPn82BAxXMM-wKHTwDBxEQFnoECBoQAw&url=https%3A%2F%2Fleeds-faculty.colorado.edu%2Fmarlattj%2Facct45405540%2FSolutions%2Fromney_ais10_im%2FIM_CH02.DOC%23%3A~%3Atext%3

 Using Artificial Intelligence to Predict and Forecast Distribution and Transportation Demands. (n.d.). Triskele Logistics. Retrieved September 25, 2023, from http://triskelogi.com/using-artificial-intelligence-to-predict-and-forecast-distribution-and-transportation-demands/