NTT DATA and Hyster-Yale deploy physical AI for manufacturing quality assurance

NTT DATA and Hyster-Yale Materials Handling (HYMH) have deployed a physical AI quality assurance system at HYMH's Berea, Kentucky manufacturing plant. The system embeds intelligence directly into an assembly workflow, using sensor data and edge computing to verify that every part is installed and each assembly stage is completed correctly before a product moves forward. Early results show the approach cuts AI deployment timelines from months to weeks compared with legacy techniques, enabling faster iteration on the factory floor.

The solution, co-developed by NTT DATA and HYMH with partner Archetype AI, integrates vision sensors, on-site edge AI processing, and advanced analytics. A physical AI model analyzes live assembly activity against expected production steps. It flags deviations immediately, allowing workers to correct issues before a product leaves the assembly area.

How the system works on the factory floor

All data processing happens locally through edge computing, so no information leaves the plant. This architecture eliminates latency and supports rapid rollout. By validating quality continuously during assembly rather than at final inspection, the model helps prevent defects from reaching downstream stages. NTT DATA designed the deployment for a critical assembly workflow, marking a first-of-its-kind use of physical AI in an industrial assembly environment.

Faster deployment and real-world impact

Traditional AI implementations in manufacturing often require extensive integration work. The physical AI model reduced deployment time from months to weeks, according to early results. That speed allows operations teams to test, refine, and scale the system across multiple production lines more quickly.

Shahid Ahmed, Global Head of Edge Services at NTT DATA, said: "This deployment shows what physical AI looks like in real production environments, not as a concept, but with tangible impact on the factory floor. By combining real production data with physical AI models at the edge, we're helping leading manufacturers like HYMH deliver high-quality products, support frontline workers and apply AI in ways that deliver real-world outcomes."

Barbara Binda, Director of Global Manufacturing Innovation at HYMH, said: "Our confidence in physical AI continues to grow, and we're starting to see the countless benefits that AI can bring to our global manufacturing operations. Working with NTT DATA allows us to leverage how physical AI can help our production teams maintain high-quality standards and deliver the most reliable products to our clients."

Why this matters for operations professionals

For operations leaders, this deployment demonstrates a practical path to embedding AI without overhauling existing infrastructure. The edge-based architecture keeps processing on-site, addressing data security and latency concerns common in manufacturing. Faster deployment cycles mean teams can iterate on quality checks more frequently, adapting to new product lines or process changes with less downtime. The approach validates assembly steps in real time, reducing the risk of costly rework or recalls and giving frontline workers immediate feedback to maintain consistent output.