In the rental industry, it can take on average 20 minutes to manually check a vehicle for exterior damage. When you consider the number of vehicles that need to be checked each day, it’s clear that such inspections can take a significant amount of time, ultimately slowing down a rental firm’s overall daily operations and costing it money.
However, recent technological developments mean these outdated, expensive, and time-consuming techniques can now be replaced with advanced automation processes, simultaneously improving performance, driving efficiency, and reducing costs.
Accurate Defect Detection
Recent advances in AI and machine learning technology mean automated vehicle solutions have become more commercially feasible in the last few years. In technical terms, this is largely because of a combination of lightweight multithreading—in which large numbers of cores are organized in groups to parallel process different workstreams—enabled by advances in graphical processing units (GPUs), with frameworks for convolutional neural networks (CNNs). These are commonly used to analyze visual imagery and provide access to cloud computing platforms that allow processing millions of images.
These technological advances allow companies to accurately detect defects on a vehicle and access the data in a matter of seconds. It means it’s now possible to inspect millions of vehicle images and accurately classify even the smallest of defects. Tiny dents, scratches, and chips that may once have been overlooked in a manual inspection can now be identified and displayed digitally in less than a minute.
For this process to be most effective, though, computers need to be able to see and interpret images as an array of pixels, meaning the resolution of those images is crucial in determining the accuracy of any defects detected. Image resolution, therefore, is critical. Given the contrast in highlights and shadows typically found on a reflective surface, such as that found on most cars today, a sensor with a wide dynamic range is needed to ensure images are of the highest possible quantity. Here, again, technical advances have made this possible.
Sony’s advanced, large format digital sensors, for example, have a much higher resolution and wider dynamic range than conventional machine vision cameras and can identify paint chips as small as just 1 mm in diameter and scratches as fine as a human hair. What’s more, large sensor sizes on the cameras mean a low level of image noise when the cars are moving.
Unlocking New Opportunities
Assessing damage to vehicles has, until recently, been a time-consuming and expensive manual process. But advances in AI and machine learning technology, coupled with powerful, high-resolution sensors, are set to change that, automating the process for greater efficiency, accuracy, and time savings. The technology, though, is still in a relatively nascent stage. Ongoing developments are set to unlock new opportunities in the years to come, and these aren’t only limited to the vehicle rental sector.
Automotive manufacturers, for instance, can deploy AI throughout the finished vehicle logistics chain. Automating inspections at the production plant and subsequently at every handover point until a vehicle arrives at a dealer will transform the process, combining improved accuracy with significant cost-saving opportunities.
Digitizing the entire process in this way, will provide an accurate record of a vehicle’s condition throughout its journey, from manufacturer to point of sale. Every partner in the chain will benefit from reliable, accurate, and consistent damage liability assignment, essential for identifying where process improvements can be made, and costs avoided. The automation enabled by AI will also allow for the redeployment of some, although certainly not all, human inspectors to more valuable functions within their business.
AI, augmented and supported by high-definition lenses with a wide dynamic range, has the potential to significantly disrupt the automotive industry for the better. The future of vehicle inspection lies in automation.
