Building the highway to AI success with cloud infrastructure

Building the highway to AI success with cloud infrastructure

Contributors: John W. Schmotzer, Automotive Ecosystem Business Development Manager, NVIDIA and Guy Bursell, Business Strategy Lead, Manufacturing & Supply Chain at Microsoft

The automotive industry continues to adopt artificial intelligence (AI) at an increasingly rapid pace to stay abreast of numerous production challenges and market opportunities. AI delivers many engineering and manufacturing benefits, from accelerating design iterations to refining quality control on production lines. AI also helps forward thinking automotive companies adapt effectively and swiftly to rapidly changing business and industry priorities.

Mapping data to the right destination

Manufacturing is the foundation of a strong automotive company, and high-quality data is the foundation of a strong AI strategy. The data landscape was once static and historical but is now also streaming and informative in real-time. Managing data of varied formats and sources is challenging and keeping data files and data sets properly labeled, merged, cleaned, and current is increasingly difficult.

But it isn’t only the pressing needs in managing large and changing data sets that test the limits of manufacturing IT teams. Changes in data usage and storage affect IT infrastructure and how well AI performs. Continuous flux in AI models, decisioning rules, and HPC projects affect data demands and processes that data feeds. In modern manufacturing, data presents more challenges beyond its ever-growing bulk, variety of formats, and seemingly endless points of origin.

“The highway to AI success is built on the integration of disparate data sets that were not traditionally combined to form new insights identified through the use of sophisticated AI technologies,” said John W. Schmotzer, Automotive Ecosystem Business Development Manager at NVIDIA.

As data management becomes a first-class priority for the automotive manufacturing community, several opportunities and challenges arise. One opportunity for streamlining manufacturing processes is the ability to merge data sets from a variety of organizations in the company to provide a single view of the vehicle across its entire lifecycle. Doing so can drive down warranty accrual costs and streamline material costs year-over-year. Warranty accrual alone on average accounts for $600 per vehicle in incremental costs for major auto OEMs and may be a deciding factor to influence customer loyalty.

New opportunities for cost optimization create new challenges for manufacturing companies. There could be a real struggle to consolidate manufacturing data, connected vehicle data, meta data, and CAE rendering files from across the company into a single view and singular format. Schmotzer dubs this the “Recursive Data Lake Problem” due to the nested insights retained in different organization and sub-organizations within the company.

Strategic executive decision making on software tools and processes that standardize upon agreed formats and ways of working will help alleviate these bottlenecks experience while pursuing industry 4.0 opportunities.

Building the highway to AI success

the advent of connected battery electric vehicles, and AI for visual inspection, we are on the cusp of a truly transformative manufacturing experience. The term “Industry 4.0″ is almost a disservice to the scope of change and real business value being realized in transportation,” says Schmotzer.

While AI is proving to be a crucial tool in meeting almost every production demand, such projects are rarely performing at peak capabilities. AI places a heavy drain on resources but modernizing a manufacturing plant into a smart factory can shrink the impact, stretch resources, optimize processes, enhance efficiencies, ensure product quality, speed R&D outputs, and lower maintenance costs.

For example, scaling-up accelerators and networks of interconnected accelerators can dramatically improve inference, AI training and model-parallel training needs, among other AI model-building and training uses. Integrated toolchains for a variety of user skill levels can leverage varying user capabilities and also democratize AI projects. Support for scaling Machine Learning Operations (MLOps) solutions is also key to higher productivity. Built-in features and metrics to ensure Responsible AI can go far in preventing future problems including potential product recalls and liabilities, among other issues.

See how Microsoft Azure and NVIDIA give BMW the computing power for automated quality control.

Deep learning (DL) poses many benefits over its less resource intensive cousin, machine learning (ML). Preserving resources consumed by deep learning projects is essential to making use of the technology sustainable. GPUs are uniquely and highly capable at resolving DL issues by virtue of their high energy efficiency and excellent price performance. GPU-powered computing is an excellent choice for heavily parallelized environments and repeatable tasks at big scales.

“Cloud infrastructure helps manufacturers break free from the chains of limitations and constraints. HPC and AI in the cloud lets manufacturers ask larger and more complex questions, the answers to which drive increasing impact, innovation and differentiation in crowded markets,” says Guy Bursell, Business Strategy Lead, Manufacturing & Supply Chain at Microsoft.

Plotting the path to execution

Pinning AI to the competitive realities in automotive manufacturing is crucial. Companies can keep up with the demand and scale of customer needs by modernizing into a smart factory, enabling optimized processes, enhanced efficiencies, assured product quality, and lower maintenance costs.

AI-first toolchains & cloud infrastructure can materially impact a manufacturer’s bottom line. Microsoft and NVIDIA’s full-stack cloud infrastructure, purpose-built for AI, delivers real-time speed, predictability, resilience, & sustainability that can help companies accelerate AI innovation and workloads.

Learn more about Microsoft Azure and NVIDIA solutions for AI:


Leave a Comment

Your email address will not be published. Required fields are marked *