AI Technology Solutions in Manufacturing

AI Technology Solutions in Manufacturing

AI has made deep inroads in several leading organisations in the manufacturing domain over the last few years. Based on the impact realised by these organisations, other organisations have been made to recognise that they need to invest in AI or they will perish. Let us examine a few examples.

BMW has implemented AI-powered computer vision systems that have resulted in reduction in false detection instances and overall enhancement of quality.

An extra layer of quality assurance is enabled as an AI solution compares vehicle data with live images of designated models to identify discrepancies, and AI-driven robots weld half a million metal studs onto vehicle frames daily and ensure precision.16 Siemens has been using predictive analytics driven by Smart Insights (SI), a platform for IIoT that provides data-driven insights for smart manufacturing. Siemens uses this platform to monitor the health of equipment at their manufacturing facilities so that downtime reduction and lower maintenance costs could be feasible.17 At their Amberg production facility in Germany, Siemens has managed to achieve a 50 per cent decrease in unplanned downtime18 and 30 per cent decrease in maintenance time at its American factory in Charlotte.19 AI-driven systems have also helped Siemens to reduce quality defects. L&T has deployed AI in predictive maintenance, leading to enhanced operational efficiency with around 10 per cent higher profitability through improved worker efficiency.20 Maruti Suzuki has managed to increase efficiency through its AI-led efforts to optimise its production lines, particularly for vehicle assembly and parts manufacturing, improving precision and reducing waste.21 AI-led preventive maintenance products have been developed by several other companies such as ASEA Brown Boveri, Honeywell, Parametric Technology Corporation, General Electric, Siemens and others that help them make their offerings more robust within their manufacturing facilities and also help their customers in critical manufacturing business domains to predict and prevent asset failures. Foxconn has implemented AI-driven visual inspection methods comprising computer vision, machine learning and deep learning in several processes related to production with the help of its partnerships with Nvidia and Google. For example, the Automated Optical Inspection system is used to inspect printed circuit boards for defects, which has helped in the reduction of inspection time by 50 per cent and enhanced the accuracy by 90 per cent. Its Small Quality Inspection system has helped in reducing defect rates in smartphone displays by 30 per cent. Foxconn also uses AI-powered X-ray inspection to analyse X-ray images of electronic components, which helps in the detection of hidden defects and reduces false positives by 40 per cent.22 AI-driven supply chain optimisation is being implemented by several manufacturers as supply chains are the lifeline of the factories. The combination of machine learning, deep learning, NLP and IoT has enabled organisations to build optimised supply chains and carry out effective demand forecasting. General Electric has been using an AI-powered system to accurately predict and streamline logistics costs that could enable a 10 per cent reduction in logistics costs. This would represent global cost savings to the wind industry of up to US$2.6billionperannumby2030basedontheindustrygrowthprojections.23TataSteelwasrecognisedbytheWorldEconomicForumin2019asoneofforty-fourLighthousesintheadoptionofFourthIndustrialRevolutiontechnologies.24AccordingtoaMcKinseyReport,theteamatTataworkingonthesuperheatingmodelpulledtogetheroperationaldatageneratedfromsensorsorrecordedmanuallyand,aftertheinitialpilot,managedtorefinethemodelandidentifiedmorethantwenty-fiveperformanceindicatorssuchastemperature,speedandothers.25Theanalyticsmodelenabledtheteamtorealise80percentofthetargetedimprovements,resultinginUS$70 to US$80 million of value and 2 to 3.5 percentage points of incremental earnings before interest, taxes, depreciation and amortisation.26 JSW Steel is the other company that has implemented AI-powered solutions successfully to monitor and control the steel-making process, optimising parameters like temperature and chemical composition for better product quality. The AI-powered platform manages close to 3,000 critical assets across ten plants and uses data from more than 13,500 sensors in real time. This has resulted in the company being able to reduce energy consumption by 10 per cent and prevent unplanned downtime as per the company annual reports.27 Hindalco, too, reduced raw material costs by 10 per cent by optimising procurement strategies with AI-led systems.28 The production facility of Bosch at Bamberg, Germany, is working towards its goal of zero defect production with the help of AI platforms with its highly automated and connected manufacturing system that is able to make terabytes of data readable in a matter of seconds.29 Procter & Gamble has managed to improve its inventory management and achieve reduction of costs by implementing its AI-driven system for demand forecasting and supply chain optimisation by working on the totality of the supply chain, including working collaboratively with retailers.30 Nike has been able to reduce inventory over-stocking due to the AI-powered sales forecasting system, which is part of its digital-first supply chain strategy.31 AI-powered robotics are being deployed for manufacturing processes by several organisations. Notable among them are automobile companies such as Tesla, BMW, Siemens and Daimler, who are using robots of their own or developed by specialist companies such as FANUC Robotics, KUKA Robotics, Universal Robotics and others for vehicle assembly, welding, inspection, material handling, painting and overall manufacturing optimization.32 Mahindra & Mahindra integrates AI with robotics in its assembly lines to improve precision and speed and reduce human error, which has resulted in 15–20 per cent increase in production efficiency.33 Bajaj Auto has deployed cobots from Universal Robots, which helped achieve zero annual maintenance costs alongside reduced power consumption.34 Use of robots in factories has not only improved efficiency and accuracy but also enhanced the flexibility required to adapt to changing production needs. AI algorithms are being used for product development for creating multiple design options. This generative design approach is being followed by Airbus to create lightweight aircraft brackets.

New Balance and Nike are using this approach to provide customised shoe insoles and shoe options to give a personalised look, feel, fit and performance.35 Application of this approach to design is increasingly being favoured by manufacturers in domains such as aerospace, consumer products, industrial equipment and automotive products.

Digital twin is yet another interesting approach towards optimisation and prediction. For this, real-world physical assets and systems are replicated virtually as a digital twin, and simulation is carried out. By blending real-time data, AI and IoT sensors, manufacturers are able to enhance reliability, optimise production costs, reduce maintenance costs and support effective decision-making. Siemens’ digital twin of a gas turbine predicts maintenance needs. GE Aviation’s digital twin optimises engine performance. The National Aeronautics and Space Administration uses the digital twin of spacecrafts for mission planning. Tesla uses the digital twin of EVs for performance optimization.36

Energy management and sustainability goals have become critical to businesses. AI-driven systems help in optimisation of energy consumption and reduction of waste. Siemens, Schneider Electric and Honeywell are some of the leading examples of organisations using AI-driven energy management systems that help with analytics regarding prediction of likely usage of energy and thus assist in the planning process to reduce wastage and environmental impact.37

Apart from these areas that are unique to manufacturing processes, the universal AI-driven systems customised to the functioning of production systems are also widely being deployed. This includes AR systems for training of production and maintenance staff, virtual assistants to handle queries from customers related to problems faced and resolution thereof, and cybersecurity to safeguard manufacturing processes from cyber-attacks.