Skip to main content

Improving Quality Inspections & Reducing Efforts

BMW Group manufactures millions of premium vehicles per year. To streamline the line-side quality assurance process of vehicle production across all brands and plants, an end-to-end industrialized solution was needed for precise quality inspections that increased inspection accuracy while reducing time, defects, and environmental impacts.

What they were looking for

An automated, robust, and highly available, quality assurance platform that was consistent in inspecting the quality and accuracy of all vehicles in production 24/7.

How we helped

Creation of a scalable and highly available Artificial Intelligence (AI) Quality Inspection (QX) platform that performs real-time visual quality inspections to catch defects and inform quality systems to resolve those defects as early as possible.

"With the integration, we were able to bridge traditional shop floor manufacturing systems with hyper modern microservice/AI services. This integration is a major strength and a key component to the success of building the Ultimate Driving Machine"

Hjalmar Hengstmann
Project Manager AI Quality NEXT

The Process

Orange Bees collaborated with the plant manufacturing IT to determine business requirements, and designed, developed, and integrated the AI and Machine Learning (ML) platform for the automotive quality inspection ecosystem within the Plant Spartanburg.

PSIA

Key Features

Extensible platform
Metadata driven use cases
Metadata subject configurations
End-to-end quality control
Document evidence
Object detection
Classifications
Dimensional measures
Anomaly detection
Semantic segmentation (color coding and labeling)
Anonymization (people and fragments, obfuscating identity)
Subject state and managing sliding confidence scale of AI models
Auto compensation in platform for line speed and other variables
PSIA

The Results

Since the implementation of AIQX, vehicle quality process has been improved by capturing and fixing defects earlier and more efficiently. The platform currently processes over 100k inspections per day, with new inspections rolling out weekly. This has helped reduce efforts tremendously, averaging around 15 seconds per inspection. The rollout of the platform started at the Spartanburg plant and has been implemented in 9 other plants world-wide, with a goal to onboard more plants and reach 2.1M inspections per day.

icon
100k
Inspections processed per day
icon
15s
Average time per inspection
icon
10
Plants implemented

Let's get started.

There as an error submitting the form. Please reach out to us directly with the contact information above.

Thank you for your message! A team member will be in touch with you soon.