By Robert L. Reid
Drones, cameras, and sensors of all kinds are making it easier for the nation’s railroads to detect potential problems with rail infrastructure. This technology is also making infrastructure inspections and improvements safer, easier, and more accurate.
Advanced technology is increasingly becoming a critical tool for the engineers who design, maintain, and work to improve the nation’s railroad infrastructure. Software-enabled devices include drones that fly above the railroad tracks carrying cameras, lidar, and other sensors to detect potential structural problems; systems that can be attached to hi-rail vehicles to gather information as they roll along the tracks; and 3D-printed fake rocks with sensors that nestle among the ballast the trains roll over, as well as other monitoring equipment placed in close proximity to a rail line.
These technologies improve safety and performance for the railroads and their employees, the employees of consultants who work around the tracks and trains, and the general public. They also offer the promise of new business opportunities. There is, for example, a pilot program for a battery-powered, autonomous railcar system that can collect important data as it moves intermodal containers individually or in trainlike platoons. It is a system meant to create greater competitiveness for railroads versus short-haul trucking while also opening up possibilities in the design of railyards and shippers’ facilities.
Avoiding hazards
Before these technologies became available, the traditional methods of inspecting railroad infrastructure for problems or gathering information to design new infrastructure involved the proverbial boots on the ground. Railroad company workers, surveyors, and teams from engineering consulting firms hired by the railroads had to be physically present, walking along the tracks, setting up instruments at specific locations, and collecting data on-site. Or they had to fly over the rail line in aircraft equipped with cameras or other devices.
But these conventional approaches posed dangers for the workers on or near the tracks, which frequently remain in operation throughout inspection and construction efforts. To keep everyone safe, the railroads often have to send people called flaggers to the jobsites to make sure work teams move back from the tracks when a train is coming and let the train crews know when to proceed.
The work of these flaggers — specially certified employees, either of the railroad itself or from third-party services — can result in expensive and time-consuming delays, explained Matthew Okubo, PLS, the president and principal licensed land surveyor for the WestLAND Group. WestLAND — a multidisciplinary firm headquartered in Rancho Cucamonga, California — specializes in engineering, geospatial, land planning, and utility-locating services, often for railroad projects.
And while the aerial inspection or surveying option can keep people out of harm’s way on the ground, it can cause other project delays because of the limited number of firms offering such services and the resulting challenges of getting the work scheduled. At railroad engineering firm Zephyr Rail, in Orange, California, one particular project faced a nearly month-long delay just to get an aerial survey conducted, plus another three months to then process the data collected, said Marc Cañas, GISP, M.ASCE, Zephyr’s executive vice president and chief operating officer.
Safety, scheduling, and costs were among the reasons why WestLAND and Zephyr turned to technology to better serve their railroad clients. For Zephyr, the answer was using drones to conduct its own aerial mapping. At WestLAND, the solution involved attaching lidar systems and various sensors to vehicles that drive either along the railroad right-of-way or directly on the tracks. In each case, these firms were following a long-standing tradition in the overall railroad industry, which “has always been on the frontier of technology,” noted Hai Huang, Ph.D., P.E., an engineering professor at Penn State Altoona. Huang is a founding faculty member of Penn State Altoona’s Rail Transportation Engineering degree program, which the university describes as “the nation’s first and only” ABET-accredited undergraduate rail degree program.
Ever since the railroads’ early influence on standardized time zones and the expansion of telegraph and telephone systems that followed the railroad rights-of-way, the industry has been a cross-disciplinary technological pioneer, combining multiple engineering disciplines, including civil and mechanical engineering and, more recently, computer science and engineering, explained Huang. At Penn State Altoona, for example, researchers are using drones to improve the safety and efficiency of railroad bridge inspections.
Such inspections can be hazardous and difficult for engineers to conduct, especially on railroad bridges that are more than 100 years old. Many of these structures were not designed with convenient access routes, platforms, or other provisions for inspectors to safely reach critical components, Huang noted.
Moreover, because many railroad bridges have remained in service beyond their originally anticipated service lives, visual inspection alone may not provide enough information to reliably evaluate their condition and structural performance.
Drones can improve access by collecting thousands of detailed photographs to create a 3D model of a bridge, while artificial intelligence can help identify and organize areas of potential concern. However, Huang emphasized that drones and AI alone cannot provide a complete assessment.
Their findings must be combined with in situ monitoring data, such as measured strain, deformation, vibration, and actual train-loading responses, to support a more reliable evaluation of the bridge’s condition and performance.
Huang added that the use of drones is intended to enhance what human safety inspectors do, not replace them.
Huang’s team has also developed SmartRock, a ballast-sized sensor used to collect data on the stability of the rock ballast that supports railroad tracks. Measuring roughly 2 in. to 2.5 in. across, versions of the device have been fabricated using 3D-printed nylon and epoxy mixed with granite powder to approximate the size, shape, and mechanical characteristics of actual ballast particles, Huang said.
When placed amid the real ballast, the SmartRock records 3D motion, including rotation and translation, and, in force-sensing versions, 3D forces generated within the ballast as trains pass overhead or maintenance activities such as tamping are performed. Current versions are battery-powered and rechargeable, including through inductive charging.
Huang’s team is also researching an “energy-harvesting” version that could generate its own electricity from the vibrations produced by passing trains and other equipment.
Penn State Altoona is also developing other rail technologies, including the use of distributed fiber-optic sensing to measure vibrations and strains along railroad tracks. By analyzing how these signals change under passing trains, researchers can assess track behavior and identify potential changes in track and support conditions — similar to how measurements such as temperature or blood pressure provide indicators of a person’s health, Huang said.
Finding a better way
About 12 years ago, Zephyr was working on a project for the proposed California High-Speed Rail when it was given a delivery schedule for an aerial mapping assignment by a conventional manned aerial mapper that would have taken months to complete. “We said there’s got to be a better way,” explained Jacqueline Patterson, Ed.D., P.E., CCM, M.ASCE, Zephyr’s cofounder and CEO.
The firm was already considering drone mapping in its workflow and looking for “a small project as proof of concept,” Patterson added. Instead, their first drone project involved mapping a section of proposed track that was 42 mi long. Zephyr managed this task by flying drones over the route in a series of sections that each measured a mile to 1.5 mi long, Cañas said.
The drone mapping effort went so well that it has become Zephyr’s specialty. The firm assembled a team of FAA-licensed drone pilots — trained in the same engineering fundamentals Zephyr teaches its own engineers as part of its workflow — and acquired six high-end data processing machines plus the necessary software. These investments mean that Zephyr can now provide services such as right-of-way condition assessments, track inspections, aerial mapping and photography, high-resolution orthoimagery (essentially, a photographic map with a uniform scale), digital terrain modeling, construction monitoring, and even community outreach to keep the public informed about the progress of work. Clients “love this because they can use the drone footage in community meetings,” Patterson noted.
The types of drones Zephyr uses are more advanced than the small devices a hobbyist might buy at a retail store, Cañas noted. They include large quadcopter units and fixed-wing VTOL, or vertical takeoff and landing, drones that measure about 4 ft across and can carry either an ultra-high-resolution camera or a lidar system, which are easily switched out. Because the drones fly at a maximum height of 400 ft, they provide images with much greater detail and density than the conventional aerial surveys, which fly overhead at much higher altitudes, he explained. The lower-altitude drone work leads to models of the site being examined with much greater levels of accuracy, precision, and clarity.
With aircraft surveys, the aerial mapping must be supported and enhanced by information from professional land surveyors who physically walk the site. The surveyors establish ground control points, paint marks on the ground, or use immobile ground features such as manholes or street corners as survey references, Cañas noted.
Although the drone work also requires input from land surveyors, “the number of points and the data collected from the field (are) reduced,” Cañas said. “So, by requiring less time spent on the ground, we are able to improve safety and schedules significantly” and provide the client with “a much richer dataset.” The goal is not to eliminate the ground surveyor’s work but to improve the process, Cañas explained.
Rolling along
WestLAND also uses drones and lidar during its land surveying and mapping work for railroads, but the lidar system is not always flown overhead. Instead, the firm developed a sensor array with lidar, GPS, and other tools that can be attached to a vehicle that drives along the railroad’s right-of-way corridor.
A railroad client liked this approach so much that WestLAND took the idea of mobile mapping one step further. “As the name implies, this mapper is compatible with anything mobile,” Okubo explained. “So we decided to mount it on a hi-rail truck and drive down the tracks to collect data.” The system eliminated the need for flaggers or other personnel physically on the ground, “which saved a lot of money and time,” Okubo noted.
The technology WestLAND uses in its surveying efforts is complex and expensive — high-end devices such as an inertial measurement unit, or IMU, can cost tens of thousands of dollars, with some systems exceeding $100,000 each. It can also be something of a “black box” technology that will give you a result when “you push a button,” he added. But whether the information it provides is always accurate remains questionable unless the user applies sound surveying practices. “To validate the integrity of the data, we utilize a series of ground control points to initially constrain each project. Additionally, we go through a rigorous process to analyze the data against unconstrained validation points to verify the information has not been distorted,” Okubo explained.
It is not unusual for large railroad projects to take a decade or more to complete, Okubo said. That means the surveyors might have to work with data collected years apart. During that period, however, the ground under the project might literally have shifted — the earth’s tectonic plates move at different rates every year, some faster than others. So, the surveyor needs to be aware of these factors and take plate movements into account over the life of a lengthy project, Okubo noted.
If a project involves any excavation or disruption of native soils, it is also critical to watch out for underground obstacles, such as utilities or other constraints that might be located at shallow depths along a railroad right-of-way, Okubo said. That is why WestLAND combines above-ground mobile mapping systems with high-speed, multi-array ground-penetrating radar units — capable of traveling upwards of 80 mph — that use sophisticated computers and algorithms to create a digital twin of the subsurface.
Introducing new technology
The Federal Railroad Administration’s engineering, technology, and automation division works to help railroads and others navigate the regulatory system to introduce new technology safely, explained Matthew Brewer, the division’s staff director. Brewer’s team of engineers and other experts conduct “a lot of engineering and risk analysis, and hazard analysis,” he said. His office also runs the FRA’s Automated Track Inspection Program, which uses both staffed and autonomous equipment to collect data on the effectiveness of track maintenance and inspection processes.
Although Brewer and his office do not endorse any particular technologies or products, he noted that drones and lidar systems have opened up new opportunities for railroads and their consultants to identify potential issues earlier, complementing traditional inspection methods. This is especially true on bridges, where the drones and lidar can collect data from locations that are difficult for people to access. While a similar investigation can be conducted through conventional inspection methods, the process may be subject to practical limitations. Thus, conducting the inspection with a drone, computer, and lidar is “cheap, simple, and accurate,” Brewer said.
Wayside detection systems — which ASCE’s 2025 Report Card for America’s Infrastructure listed as one of the railroad industry’s “solutions that work” — help railroads collect critical information about their trains. For example, the combination of high-resolution cameras and scanners located around a railyard, plus the installation of radio frequency identification systems on railcars, enable railroads to track critical details about components on individual pieces of rolling stock, Brewer noted. Weeks or months later, when that same railcar returns to the same yard, the AI-powered system will help the railroad identify any changes to the railcar, its wheels, or other elements.
This data “lets you match up the current inspection with prior inspections to look for subtle differences,” Brewer noted, and can be used “for predictive analytics or to show that something is starting to fail.”
Brewer’s team can also study new technologies under a condition called “shadow mode,” in which the tech is introduced unofficially on a limited basis. Currently, one such shadow mode project involves sensors that count the number of axles that cross two nearby points at a gated grade crossing. The system is designed to keep the crossing gates closed until all the axles recorded by the first sensor have moved past the second sensor, indicating that the train has cleared the area.
Such systems have been used successfully in other countries but have not yet been implemented in the United States, Brewer said. That means there is no U.S.-based safety case study yet, which is a requirement for regulatory approval. The shadow mode approach is intended to try out the technology and help prove that it can be used here safely, Brewer explained.
Testing tech
Brewer also mentioned a more official FRA pilot test of a new technology that could help railroads improve their operations, find new business, and design new facilities.
Matt Soule, who has an aerospace industry background, is the CEO and cofounder of Los Angeles-based Parallel Systems. The firm has designed an autonomous, battery-electric railcar system that can carry individual container units from place to place or combine railcars in a platoon of as many as 30 autonomous rail vehicles that can travel together in close proximity without being coupled. The system is being tested with FRA approval along a 160 mi stretch of rail line in southeastern Georgia.
And while the Parallel system is designed to operate on the main railroad lines, it does offer the opportunity of expanding the rail infrastructure. Because the battery-powered Parallel system — which weighs about 40,000 lbs unloaded and 140,000 lbs when loaded with a container — is much lighter than a 400,000 lb diesel locomotive, the engineers who design rail spurs and rail-served inland ports would no longer have to plan the infrastructure to accommodate the weight of a full train consist, Soule said. Instead, they could use lighter rail gauges, fewer ties, and less massive bridges, as well as making other changes, he explained.
Likewise, Parallel would enable a railroad to work with a freight terminal that has a smaller footprint — perhaps only 10 acres rather than 200 acres — or grade elevation changes too steep for a full-sized train, Soule said. It would also be easier to construct a lighter-gauge spur line directly to a shipper’s facility, making it a small terminal, he added. Such changes could help the $80 billion rail industry better compete with the $800 billion trucking industry, especially for short-haul operations, Soule said.
Because the autonomous rail system operates with various cameras and sensors, the technology could also help identify any problems or potential hazards along the rail route and then alert maintenance teams to take a look, Soule concluded.
With the array of new technologies available, the rail industry can continue to provide its traditional services while also moving in new directions in safer and more innovative ways. The tracks ahead look clear.
Robert L. Reid is the senior editor and features manager of Civil Engineering.
This article first appeared in the September/October 2026 issue of Civil Engineering as “Rolling with Rail Tech.”
