AWS Transfer Family

BMW Group uses AI-powered vision and sound analytics for predictive maintenance and quality management

The BMW Group built its Vision and Sound Analytics Service (VSAS) on AWS, processing 1.3 million image and audio files daily across 1.3 PB of data at more than 16 production sites worldwide. The company uses the platform's stored data to build AI models for predictive maintenance and quality management, including a computer vision model that predicts potential rail failures to reduce downtime, and an Acoustic Analytics application that automatically captures and analyzes vehicle sounds using trained models to detect anomalies and improve engine performance and in-cabin comfort, reducing the need for human intervention in testing vehicles with no driver present. Using AWS Graviton processors for AI model training, BMW cut compute costs for its computer vision solution by 63% and reduced model training time for various use cases from days to hours.