Many experts agree that AI will have the most significant impact on manufacturing. According to McKinsey Research, AI can create $1.2 Trillion to $2 Trillion of value in supply-chain and manufacturing. Manufacturing processes generate enormous amounts of data, involve repetitive tasks, and present multi-dimensional problems beyond the scope of many conventional tools. The industry is also projected to face a workforce shortage due to skilled employees’ looming retirement. AI and Automation are key technologies that can address this gap while increasing operational efficiency, improving quality, and boosting productivity. However, AI has yet to gain significant momentum and reshape manufacturing. Manufacturing executives and plant leaders must overcome several challenges before AI-led digital transformation transitions from a select few to a broad market at scale :
- Legacy Infrastructure – The production sites typically have a wide variety of machines, tools, and systems that use disparate and often competing technologies. For example, discrete manufacturing is dominated by industrial robots from Kuka, ABB, and Fanuc that require different skills to integrate, program, and troubleshoot. Industrial engineers often struggle with connecting legacy machines as they have to carefully think through connectivity options and find the right Industrial IoT gateway compatible with Ethernet, EtherCAT, MQTT, OPC UA depending on the machinery used at the plant. The burden usually falls on system integrators to find sensors, protocol converters, and gateways they need to install. An ecosystem of players that can offer compatible components that use standard rules and frameworks to connect to ERP, MES, and PLC/SCADA systems becomes critical. Without universal connectivity, there is no way to collect data, AI, and ML applications.
- Access to High-Quality Data – Data is spread across multiple databases in multiple formats, not suitable for analytics. Many manufacturing companies lack data infrastructure or do not have enough volume or quality data. Data quality and data management issues are critical, given the high reliance on quality data by AI and ML projects. A Predictive Maintenance solution will need data from a computerized maintenance management system or process historians and may require database connectors or custom scripts. The manufacturing site may also be remote, bringing additional complexities in terms of data storage. Depending on the security policy, the IT team may also not want to send data to the cloud, requiring on-premises solutions. However, the traditional approach companies take to solve any data issues requires months of effort. This time and the initial efforts associated with this approach often cause projects to fail after only a few months of investment and work. AutoML 2.0 platforms that provide AI-focused Data Preparation, Feature Engineering automation, ML automation, and automated production are the future of data science for data-driven manufacturers. End-to-end data science automation involves automating the entire data science process and addresses data preparation issues and renders faster delivery of insights from months to days.
- Real-Time Response – Many manufacturing applications are sensitive to latencies, such as predictive maintenance or predictive quality, and require an ultra-fast response. For these applications, the system cannot wait for the round trip journey to the cloud to perform data processing and get actionable insights. The decision has to be made in real-time, acted upon immediately in a few milliseconds. It becomes more efficient to process data locally near the source of data for faster response. Real-time data processing allows manufacturers to take action immediately and prevent undesirable consequences. For example, by using predictive analytics for quality, manufacturers can identify defective components and perform rework or replace the faulty part preventing a product recall. Real-time decision making and ultra-fast response time require edge-based computing architecture. The ability to deploy predictive models on the edge devices such as machines, local gateway, or server is critical to enable smart manufacturing applications.
- Transparency & Bias – The AI technology stack is extraordinarily complex and challenging. People without a data science background struggle to understand how predictive modeling works and do not trust the abstract algorithms behind AI technology. Transparency implies providing information about the AI pipeline – the input data used in the process, algorithms selected, and how the model made predictions. One approach to increase trust is to provide details about the AI workflow. That includes providing a detailed process to transform the raw data into machine learning inputs (a.k.a. feature engineering), and how the ML model produces predictions by combining hundreds of or even more features. Lack of transparency is where Explainable AI can help. By giving insight about how the prediction models work and the reasoning behind predictions, manufacturing organizations can increase transparency and remove the bias against AI.
- AI Readiness & Technology Culture – AI projects require multi-disciplinary teams with expertise in data management, algorithms, and machine learning. Most OT experts are not trained in building predictive models or using advanced analytics powered by AI and ML. Instead, they rely on data scientists and advanced analytics teams to assist them in using the latest technologies in operations. How can manufacturers hire AI and ML talent when it is in such short supply? The perception that the manufacturing industry is risk-averse, not sexy or cool makes it even harder to attract AI talent. The answer may be leveraging AutoML tools. Finally, manufacturing needs to embrace the tech culture to move fast, iterate rapidly, and make tech-driven innovation core capability.
Manufacturing CIOs and analytics leaders should look at AI/ML platforms to accelerate digital transformation initiatives. AI automation can reduce the cost to implement AI and also speed up the painfully slow AI deployment. By using AutoML tools, SME’s can focus on day-to-day responsibilities while automated data pre-processing and feature engineering will enable them to build predictive models at the click of a button. Explainable AI and transparent features create trust and will garner buy-in from domain experts. The new AutoML 2.0 platforms automate up to 100 percent of the AI/ML development workflow, using an AI-based engine to automatically discover meaningful patterns and build ML-ready feature tables from operational data.
To learn more about AI/ML workflow, top challenges, and how to overcome them, read our white paper on Practical Challenges of Enterprise AI.