Rapid advances in computing power and artificial intelligence provide the digital tools to automate the inspection process and more accurately detect defects. By Benjamin Jones
In today’s world, disruptive technologies such as artificial intelligence (AI) and IoT are transforming automotive operations. Global automotive giants such as Toyota, Jaguar Land Rover and Ford are already leveraging automation to replace manual inspection processes throughout the finished vehicle supply chain. These companies are embracing it as part of Industry 4.0: using automation, machine learning, and real-time data to drive meaningful business benefits. The outdated manual processes of traditional vehicle inspections, which are not only costly but also time-consuming, are now being replaced with advanced automation processes to simultaneously improve performance, increase efficiency and reduce costs.
The role of AI and machine learning in automating vehicle inspection
AI and machine learning have a long history, but it’s only very recently that advances in hardware and software have made automated vehicle inspection solutions commercially feasible. The recent combination of lightweight multithreading (large numbers of cores organized into groups to process workflows in parallel) made possible by advances in graphics processing units (GPUs), combined with frameworks for convolutional neural networks (CNN ) and easy access to powerful cloud computing platforms to process millions of images, enables companies to accurately detect faults on a vehicle and deliver the data in seconds.
Artificial neural networks reproduce the functioning of the human brain; the nodes are connected by a network of weighted links, equivalent to the biological nodes of the brain which are connected by synapses. Whereas in conventional computing data is stored in bytes, 0 or 1, and modified by Boolean operators (e.g. addition, subtraction, etc.), for neural networks it is the weight of each link between two nodes which is important. A neural network has multiple nodes, creating a pattern of links between them, which makes it possible to model even extremely complex features.
Therefore, by feeding large amounts of previously seen and annotated data, these features can be modeled by the neural network and, in turn, used to understand if the same features are present in new unseen data. These techniques have been used very effectively in medicine to train neural networks to recognize diseases and in turn process large sets of data that no human could reliably or efficiently understand, diagnosing patients in much more precise and consistent than before.
GPUs are perfectly suited to speed up this process because they perform a large number of multiple and simultaneous calculations, unlike traditional computer chips, to model the links in the neural network. Access to scalable computing power in the cloud provides the third component: the means to process and store massive amounts of data without having to build expensive infrastructure in-house or on-premises.
Pinpointing precisely where these damages occur in the value chain enables the correct allocation of responsibilities, a prerequisite for improving processes
These technological advances now make it possible to inspect millions of vehicle images and accurately classify even the smallest defects. Dings, dents, scratches and chips that would often be overlooked by an inspector can be identified and displayed digitally on a dashboard in less than a minute. Consistent results are achieved in every site and every inspection, a huge advantage for an OEM looking to standardize quality across a global supply chain. This visible damage is a significant cause of warranty costs for OEMs and liability for damage in transit for logistics providers. Pinpointing precisely where these damages occur in the value chain enables the correct allocation of liability, a prerequisite for process improvement.
The benefits don’t stop there. Trained algorithms are also able to identify specification errors in the vehicle assembly process. Incorrect parts or even missing components, such as spoilers, wheels and badges, can be reported in seconds, allowing them to be reworked before a vehicle leaves the factory.
Empower AI with high resolution images
Computers must be able to see and interpret images as an array of pixels. The resolution of the vehicle image is therefore critical in determining the accuracy of fault detection. In short, image resolution is everything. Due to the contrast of shadows and highlights on a highly reflective surface, a sensor with a wide dynamic range, or gamut, is also essential to ensure a high quality image. And again, technical advancements have made this possible. Sony, for example, provides the most advanced large-format digital sensors on the market, capable of identifying scratches as fine as human hair and paint chips as small as 1 mm in diameter. These sensors are also extremely durable, with much higher resolution and a wider range than conventional machine vision cameras.
Lighting is the other key factor needed for the highest resolution images. Automated vehicle inspection specialist DeGould, for example, deploys a patented combination of three different light fields in the Self-compact system: white light, black light and band structured light, which are specifically designed to meet the different spectral properties of different types of defects. For example, many scratches and dents will change from highly visible to entirely invisible depending on vehicle color, light output, and body panel spectral properties.
The future of car inspection
Today, the real game-changing opportunity for OEMs is to deploy AI throughout the finished vehicle supply chain. By automating inspections from the production plant and at each point of drop until the vehicle arrives at the dealership, the process can be transformed. Improved accuracy can be associated with significant cost reduction opportunities.
Advances in technology have made this transformation both feasible and cost effective. By digitizing this entire process, a one-stop solution digital vehicle passport can be created, providing an accurate record of the vehicle’s condition throughout its journey. All partners in the chain will benefit from a correct, consistent and reliable allocation of liability for damages, an essential first step in identifying process improvements and avoiding costs. Automation also allows some, but certainly not all, human inspectors to be redeployed to higher value functions within the company.
About the Author: Benjamin Jones is the Head of Machine Learning Strategy at DeGould