The long-distance trucking industry faces several significant problems, and finding safe overnight truck parking spaces has remained one of the most challenging. Given the vital role of trucking in the U.S. economy, making it easier for truckers to locate open parking spots should be a top priority for trucking companies and state transportation departments. As technology rapidly develops, computer vision software coupled with live camera feeds of parking facilities can help truckers with this struggle.
The use of trucks to move freight in the U.S. is growing and saw a 27% increase in ton-miles moved from 2021 to 2022, the latest data available from the Bureau of Transportation. According to the Federal Motor Carrier Safety Administration, 3.36 million combination trucks are registered in the United States. However, in U.S. public rest areas, there are only 40,000 truck parking spaces.
Improving access to truck parking space information would help alleviate some of the system-wide stress. Truckers often rely on third-party apps to plot their routes and to find available rest areas during long hauls. One popular app, Trucker Path, accomplishes this using crowdsourced data similar to popular navigation apps like Waze. The more users, the more accurate the data. Trucker Path receives updates from “[600,000] truckers” on parking availability to help other truckers plan their routes. The app also estimates the likelihood of open spaces based on parking history at a given location.
While this service is helpful, there’s room for an alternative that receives data directly from the source—a live feed displaying parking availability that doesn’t depend on truckers using a particular app. Some states have conducted pilot programs on parking information feeds, but in 2017, the Kansas Department of Transportation (KDOT) tested combining computer vision with camera feeds.
In 2017, KDOT began work on a truck-parking information management system and started monitoring 160 truck parking spaces at 16 public sites. KDOT’s proposal used camera poles installed near the truck parking areas to create a 3D model of the parking area and send updates to signs along the Interstate, showing parking availability for the next two to three rest areas ahead.
Given recent technological advancements, this system could be further improved. Pattern recognition software can be trained to detect parking spot openings with only one camera, so long as the parking spots are in view. There’s less of a need for a 3D model of the parking area itself, and camera images have increased in resolution since 2017. Additionally, in rest areas where cameras already cover much of the designated truck parking, implementation would be cheaper.
Currently, parking availability is tracked in several ways, including parking garages, rest areas, and other facilities. Conventional parking availability systems use schedules, crowdsourcing, or in-ground sensors installed at each parking space to determine whether a spot is occupied. The latter method of in-ground sensors often comes with high costs for both installation and maintenance.
In contrast, software can incorporate multiple data sources from in-ground sensors, camera feeds, schedules, and so on to paint a more comprehensive picture of parking availability. Computer vision object-detection algorithms can analyze camera feeds to detect whether a space is being used. Given the relative costs of in-ground sensors and high-resolution cameras, leveraging computer vision and cameras can greatly reduce costs compared to traditional in-ground parking availability systems.
Many parking facilities already have security cameras that could be leveraged by computer vision systems, reducing initial setup costs. And those that don’t could install them, improving security and parking management.
Figure 1 provides a visual example of how such a system would operate. Cameras detect a car’s license plate upon entry into the lot and as it moves through the garage. When the car parks in a space, the “condition” of the parking spot is changed from zero to one, indicating that the spot is filled by the identified car, in a simpler system focused strictly on availability, the license plate number would be irrelevant and could be cut.
Figure 1: Parking Detection Example
Source: In Hwan Jung et al., “Smart Parking Management System Using AI,” Webology 19 (2022). https://www.webology.org/data-cms/articles/20220123012239pmWEB19307.pdf (24 Apr. 2024).
The data could be fed to roadside LED displays outside rest areas or other truck parking locations. It could also be made freely available through a public application programming interface (API) so that any driver app could fetch real-time parking occupancy data.
Since state transportation departments run rest areas, this approach could be implemented via a pilot program at the state level, similar to KDOT’s approach in 2017. Improved parking information would enable better allocations of scarce parking spaces and reduce time wasted by drivers searching for parking. This would enhance the safety of trucks and the efficiency of their freight movements.
If a program like this is successful and lowers costs, other states could work to emulate its success as a promising alternative to other existing methods of parking availability tracking.
Computer vision software coupled with live camera feeds of parking facilities is a promising alternative to conventional static parking availability signs. State governments should examine and implement cost-effective technologies that could help truckers find the best places to park.