Case Study: Elektronische Fahrwerkssysteme achieves large-scale deep-learning road analysis for autonomous vehicles with Microsoft Azure

Audi technology partner EFS uses deep learning to analyze roads for self-driving vehicles

Elektronische Fahrwerkssysteme GmbH (EFS), Audi’s lead chassis-development partner in Gaimersheim, Germany, faced the challenge of giving automated vehicles a human-like, anticipatory understanding of roads from high-resolution 2D camera images—detecting road boundaries, relative distances, and hazards even when parts of the scene are occluded or ambiguous. EFS needed a scalable proof of concept to show deep learning could handle these large, complex image datasets before moving to product development.

EFS built a solution on Microsoft Azure using NC‑series VMs with NVIDIA Tesla P100 GPUs, Azure Blob and Disk storage, and proprietary recursive algorithms that analyze progressively lower-resolution versions of images to reduce compute time. They generated labeled training data via a simulation “game,” leveraged Azure’s scalability and storage integration, and validated the approach: the deep-learning tests succeeded, enabling EFS to advance product development and marking one of the first large-scale demonstrations of this technique.

Elektronische Fahrwerksysteme

Max Jesch

Software Developer