Until now, it has not been possible to accurately predict the time it takes to transport freight across the USA-Mexico border, which can be an important factor for cargo shippers. Researchers at the Texas A&M Transportation Institute (TTI) have now found a mathematics-based solution to the problem.
Though careful vetting of shipments crossing the border is necessary to maintain security, unreliable border wait times can cause major slowdowns for freight. In 2008, the TTI demonstrated a system to the Federal Highway Administration (FHWA) that accurately and reliably used technology to collect border wait time data. Over the course of several research projects, TTI researchers created a solution that uses radio-frequency identification (RFID) technology, currently present in most trucks, to measure border wait times. Deployed at seven commercial ports of entry across Texas, this system provides anyone interested, and especially US Customs and Border Protection agencies, with reliable estimates via the website TTI created.
Before the website was available, shippers relied on the free travel-time estimates provided by Google to predict cross-border travel times, but the time spent at the crossing was not included in Google’s estimate. TTI’s latest approach uses the border wait times from the website combined with travel times from Google to provide a better travel-time prediction. The new system relies on an algorithm developed by Jose Rivera Montes De Oca, a Texas A&M University graduate student studying mathematics. Beginning with data generated in 2013, the algorithm takes reams of historical data and predicts the expected wait time at the border for a given day and time.
The new algorithm now updates the TTI-developed website’s estimate 48 times a day, providing shippers with a highly-accurate travel-time estimate for commercial vehicles passing through border checkpoints. This results in more efficient cross-border supply chains, and that can mean a better bottom line for freight shippers and, potentially, savings for consumers.
“We’re supplying that missing piece of the puzzle,” said TTI software developer Swapnil Samant. “With the machine-learning algorithm we developed, we can predict accurate travel times from origin to destination and post them to the website for a given 24-hour period. And we refine the estimate every half hour. It wouldn’t have been possible without Jose’s expertise.”
De Oca commented, “Sometimes the field of mathematics is so theoretical, you can’t really explain what you do to other people. But I can point to the website and show them how my work makes a difference.”