Artificial Intelligence (AI) is rapidly transforming almost every industry. For the manufacturing industry, AI is revolutionary. Historically, manufacturing has been heavily codified, relying on established processes and methods developed over time. However, with technology has come the potential to transform performance across the breadth and depth of manufacturing operations. AI is the perfect fit for the manufacturing industry; from upgrading product development to heightening the capabilities of quality control – potential applications are just shy of limitless.
AI in manufacturing
In order to harness its potential, manufacturers must be deliberate with how and where they apply AI, to achieve the most lucrative results. Capgemini’s research highlights that 51% of top global manufacturers in Europe are implementing at least one AI use case. However, while focusing on the most promising use cases, manufacturers must anticipate the next step and be prepared to overcome the hurdle of scaling AI across their operations.
The best use cases for AI in manufacturing
For manufacturers embarking on a transformation journey, there are critical benchmarks to achieving success when deploying AI. According to recent research by Capgemini, which analysed 300 leading global manufacturers, there are three use cases that stand out, to initiate an AI transformation. These are: intelligent maintenance, product quality control and demand planning. These cases are the ideal places to start, due to a combination of common characteristics, which include:
- Clear business value/benefits: focusing on use cases where benefits are easily identified and quantified, including in financial terms, makes building the business case easier
- Ease of implementation: focusing on less complex use cases leads to shorter payback periods, as well as a higher return on investment (ROI), which further strengthens the business case
- Availability of data: selecting use cases with a wealth of actionable data, will increase the functionality of AI systems.
Intelligent maintenance has shown to be the “low hanging fruit” of AI adoption; beyond minimising downtime, AI-enabled intelligent maintenance also reduces maintenance costs and increases productivity. Similarly, in-line product quality control presents a good opportunity for AI implementation, with the potential leverage new sources of data, with minimal effort. A case in point is Audi, which has installed an image recognition system based on deep learning at its Ingolstadt press shop. Several cameras installed directly in the presses capture images of pressed sheet metal, which are analysed by the AI system to identify any faults. The system was trained using several million test images and has managed to achieve a very high accuracy. Lastly, organisations are using machine learning to predict changes in consumer demand. Based on historical consumer data, organisations can make the necessary changes to production schedules and raw material procurement. Better forecasting yields several benefits from better client service to inventory reduction.
Moving from POC to AI at scale
AI in manufacturing operations has huge potential to reduce costs and improve efficiency, however, as with any new application of technology, challenges remain. Most commonly manufacturers struggle to scale their experimentations beyond the proof of concept (POC) stage. Drawing on our experience of working with organisations to scale AI experimentations, Capgemini has been able to identify key steps that manufacturers can take to scale successfully.
The first step in achieving scale involves bringing the AI prototype up to speed with processing data in real time from the shop floor/production environment. To automate the collection of real-time, live data, the prototype needs to be integrated with legacy IT systems. Legacy manufacturing systems – such as enterprise apps for product lifecycle management, manufacturing execution systems (MES), and enterprise resource planning (ERP) systems – have multiple data sources. In order to be effective, AI applications require data-rich sources. In addition to these, AI systems will sometimes need more granular data. This would come directly from machines and equipment, such as IoT systems.
Put down solid foundations
To create a robust foundation for scale, and to encourage new implementations, manufacturers should design a data governance framework – or set of guidelines – that defines critical processes related to the generation, management, and analysis of data. In addition, they need to deploy a data and AI platform – a unified platform to store and analyze data using AI and to make it available to issue-specific AI applications. This is crucial, to ensure organisations are continuously acting upon real-time feedback, maintaining a high standard of accuracy.
Alongside governance and platforms, talent will also be a key building block, including manufacturing-specific expertise in AI, data science, and data engineering. Investing in foundational technology and AI skills allows organisations to maintain momentum when the value of AI has been proven by the first few use cases. It also helps in creating repeatable, faster, and easier rollouts of new AI applications in the future.
Scale the solution across the network
Once the AI prototypes have proven their value, they need to be scaled to the plant level and across the broader manufacturing network of the organisation. This is achieved by building on the key foundations of data management and talent that have been set in place. Once organisations have developed their data and AI platform, existing AI implementations can be transferred to the platform, to leverage the full value of available data and resources. Building on these foundations, manufacturers will establish a culture of insights-driven operations, where AI progressively takes over routine tasks, freeing up management attention from running day-to-day operation towards the development and implementation of longer term optimisation strategies.
With such strong potential to revolutionise manufacturing operations, it is no wonder that adoption rates are increasing. However, while we find that major global manufacturers have started experimenting with AI use cases, scaled deployment is rare. Unless more organisations move from pilots and proofs-of-concept to scale, then a new 4.0 era in manufacturing will remain an elusive goal. By adopting a scale-driven strategy – which focuses efforts on the most valuable use cases and lays down strong governance, platform and talent foundations – manufacturers can turn the revolutionary potential of AI into a reality.
The author, Pascal Brosset, is Chief Technology Officer, Digital Manufacturing, at Capgemini.
Back to Homepage
Back to Technology & Innovation