RyderVentures, the venture capital arm of Ryder System, last year was one of the companies that collectively invested $17 million in Terminal Industries, which was developing an AI-based system for monitoring a myriad of yard applications. That $17 million eventually turned into $31 million in seed funding.
Now, Ryder and Terminal say they have achieved a milestone: a high rate of monitoring accuracy in a pilot program at a Ryder Supply Chain Solutions (SCS) facility in City of Commerce, California.
In a prepared statement Wednesday, they said the pilot was designed to “leverage Terminal’s computer vision technology to automatically index and analyze trucks and trailers flowing in and out of the warehouse yard.” The City of Commerce facility is part of the e-commerce operations of SCS, one of three segments at Ryder (NYSE: R) along with its flagship Fleet Management Services (FMS) and its Dedicated Transportation Solutions.
The pilot has processed more than 10,000 “truck detections.” Detections involve the system using computer vision and its affiliated machine learning and AI to record such things as license plates and Department of Transportation and trailer identification numbers, and to monitor adherence to schedules, according to the statement.
But the camera-based computer vision technology process needed something to compare itself to, so clipboard and paper processing continued during the pilot. Ryder said the program reached 99% accuracy when compared to the information produced by the manual check-in and monitoring system.
Michael A. Plasencia, group director of new product strategy at Ryder and managing director of RyderVentures, said Ryder had long believed it should pursue automation of its terminal yards with a development partner.
“They had this technology and they had this use case and platform,” Plasencia said of Terminal in an interview with FreightWaves. “How do we make it work for Ryder?”
Strategy at RyderVentures is not similar to a venture capital company that is, as he said, “purely financial-based. For us, it’s leveraging the investment arm to really develop deeper strategic relationships with the founders, giving these technologies a platform to test and perfect their products and work with them.”
The plan is to “really work together to get it to where it needs to be,” he added. “So it’s been over five months and during that process, we’ve seen a great start.”
Work on the pilot began in January, according to a Ryder spokeswoman, but the cameras were turned on in March.
One of the philosophies behind Terminal, CEO Max Constant said in the shared interview with Plasencia, is that “every major problem or a major opportunity or use case in the yard, regardless of what kind of yard it is, comes down to understanding the movements of trucks, trailers, chassis and to a certain extent people.”
The longer-term goal would not be just to provide an AI-based system that replaces the clipboards of an individual yard, Constant said. “There’s a moment where the technology is such that you can connect yard A, yard B, yard C to infinity.” He called it “the language of the yard.”
“Now you can future-proof the yard in such a way that scales almost indefinitely.”
Not surprisingly, the existing clipboard-based system “has a lot of room for error,” according to Plasencia, involving not just taking down information in the yard but also keying it into a database. “So this was just a pretty clear case where we know if the technology would work, it was going to be adding value,” he said.
The program has been extended to an FMS yard in Dallas. However, whereas the City of Commerce pilot involves stationary cameras taking pictures of vehicles, in Dallas, the Terminal software is used to survey the inventory of vehicles for rent or lease on the lot, a process that normally is extremely time-consuming.
Plasencia said the existing system involves regular clipboard-based “yard walks” to count the inventory on the site. But the Terminal software can be used on a tablet or cellphone. And while the yard walks are still necessary, the “walker” scans the properties on the lot and the software does the count.
Yard checks are “very manual,” the Ryder spokeswoman said. “They happen daily, often multiple times a day nationwide.”
Constant said Terminal’s philosophy is to “start with the hard scientific problems and work backward.” The idea is to “build the science, and then the applications are a derivative of that.” That contrasts, he said, with other computer vision-based solutions that “start with the application and then figure out the science.”
Taking images of trucks from the internet and downloading them into a program is not adequate to develop an algorithm that can deal with “mud or occlusions or harsh lighting,” he said, referring to common yard conditions.
But linking up with a company like Ryder “meant we could bring the science as close as possible to the actual data itself so that we could get those algorithms to a point where they were reliable at 99%,” he said.
Asked about cost, Constant declined to provide specific numbers on configuring a yard with computer vision software. However, the goal is to “deliver 10 times the value at a tenth of the cost,” he added.
The cameras are off-the-shelf equipment. The job at City of Commerce required only four cameras, and Constant said the company would seek to get that number down to two.
With two of the three divisions at Ryder operating a pilot – SCS in California and FMS in Dallas – that leaves the Dedicated division on the outside looking in. But the Ryder spokeswoman said the company is looking at use cases in that group, “so we do see opportunity across the business.”
Among the other investors in Terminal are supply chain investor 8VC; Prologis Ventures, an arm of warehouse operator Prologis (NYSE: PLD); and NFI Ventures, an arm of privately held truckload carrier NFI.
