Amazon has introduced an innovative Artificial Intelligence (AI) technology, which can identify the most minor irregularities in delivery vehicles, such as tire deformations, undercarriage wear, or bent body parts, well before they pose on-road issues. This cutting-edge system, known as Automated Vehicle Inspection (AVI), brings peace of mind to fleet managers who previously depended solely on manual inspections and human observation for their daily safety checks. Amazon is rolling out the AVI technology in collaboration with the tech startup UVeye in the United States, Canada, Germany, and the United Kingdom.
My utmost concern is avoiding preventable incidents, such as a tire blowout due to an unnoticed defect in our morning inspections. This technology enhances the safety of our fleet.
Bennett Hart – Owner of the Logistics Company Hart Road
The enhanced safety for approximately 280,000 drivers employed by Delivery Service Partners (DSPs), such as Hart Road, across the globe, who assist in delivering packages to Amazon customers, is just one of the technology’s advantages. Another notable benefit is the scalability of AVI, which is particularly valuable considering that DSP-affiliated drivers deliver a staggering 20 million packages to Amazon customers worldwide daily.
One compelling advantage of AVI is the comprehensive insights it provides to fleet managers. It can monitor identified vehicle issues and determine if they are recurring on specific routes.
Tom Chempananical – Global Fleet Director at Amazon Logistics
How AVI works
After each route, drivers pass through an AVI arch, traversing a series of plates outfitted with sensors and cameras.
When you visit the doctor, you anticipate getting a scan; we essentially perform a similar function but for vehicles.
Amir Hever – CEO of UVeye
As the vehicle moves at 5 mph, the AI system conducts a complete vehicle scan within seconds, swiftly detects issues, categorizes them by their level of severity, and promptly transmits the results to a computer. Subsequently, a Delivery Service Partner (DSP) can assess the necessary repairs and maintenance to ensure their vehicles are roadworthy for the following day.
It can detect everything, And it’s virtually instantaneous.
Bennett Hart
Although the technology was initially developed for inspecting the undersides of vehicles at border crossings and security checkpoints, it has evolved to utilize AI for scrutinizing finer, more intricate vehicle details, including damage. AVI relies on “machine stereovision,” a technique that employs two perspectives to create a comprehensive 3D image. It also utilizes deep learning, a branch of machine learning that employs layered neural networks to replicate the learning mechanisms of the human brain.
“We couldn’t just pick up the UVeye solution from the shelf and request DSPs to adopt it,” Chempananical emphasized. “Acknowledging the distinctive requirements of DSPs with fleets exceeding 100,000 delivery vans worldwide, we collaborated directly with UVeye to educate the AI models and algorithms according to Amazon’s stringent safety standards to ensure the smooth operation of our vehicles.”
A game-changer for fleet maintenance
AVI has revolutionized the inspection process by making it faster, more precise, systematic, and impartial. Already, it has unveiled concealed damage patterns, such as the discovery that 35% of all problems are tire-related. These issues encompass sidewall tears, debris, and nails embedded in treads—problems that were challenging to identify through manual inspections. AVI enables DSPs to receive timely alerts for tire replacements before they escalate into major concerns. This proactive approach not only enhances safety for drivers and other road users by averting potential tire blowouts or flats but also eliminates potential delays for customers.
The wonderful aspect of AI is that every instance of damage is subsequently integrated into an application programming interface (API), which refines models and enhances detection accuracy. In essence, the more AVI is employed, the more it improves.
Chempananical
We conduct the analytical processes on Amazon Web Services (AWS), where we process and store the extensive volume of vehicle images and data, measured in terabytes. Here, we accomplish this using Amazon Simple Storage Service (Amazon S3), Amazon Elastic Compute Cloud (Amazon EC2), AWS Lambda, as well as other related services. We generate the results in no time and disseminate them via the API. Amazon has seamlessly integrated these results into the current DSP interface, presenting fleet managers with detected issues, complete with accompanying photos and recommended repair actions.
Comprehending the intricacies of delivery fleet tires is merely the beginning. This digital microscope is capable of monitoring various other particulars, such as detecting potential trip-and-fall risks on the cargo steps and identifying damage to hazard lights.
This technology has the potential to establish itself as the new benchmark for vehicle inspection. Amazon consistently focuses on scalability, and our collaboration with one of the world’s largest fleet networks is facilitating our expansion. This partnership serves as a constant reminder of what we have created and how to effectively implement it in numerous delivery stations, dealerships, and other settings.
Amir Hever
