Honda and ODOT test vehicle-based system for automated roadway inspections

Honda and the Ohio Department of Transportation have completed a two-year pilot program testing an automated roadway inspection system that uses vehicle-based sensors to identify road and infrastructure deficiencies. The project, led by Honda and funded by ODOT through its DriveOhio smart mobility initiative, evaluated whether data collected from production-style vehicles could supplement or partially replace traditional, labor-intensive roadway inspections.

The pilot focused on a prototype Honda Proactive Roadway Maintenance System, which relies on a combination of onboard cameras, LiDAR sensors, and artificial intelligence models to detect common roadway issues. These include potholes, faded or missing lane markings, damaged guardrails, obstructed or degraded road signs, rough pavement surfaces, and hazardous shoulder drop-offs. The goal of the system is to provide transportation agencies with timely, actionable data that can support more proactive maintenance decisions while reducing costs and improving worker safety.

Testing was conducted across approximately 3,000 miles of roadway in central and southeastern Ohio. According to project documentation, the routes covered a mix of rural and urban environments, multiple road classifications, and a wide range of driving conditions, including different weather scenarios and times of day. ODOT personnel operated Honda test vehicles equipped with the sensor suite, allowing the agency to directly observe how the system performed under real-world conditions rather than controlled test environments.

The pilot was carried out in collaboration with several partners. i-Probe Inc. contributed data validation and analysis expertise, particularly for road roughness and lane marking conditions. Parsons Corporation provided systems integration support, including the use of its iNET Asset Guardian platform to translate raw detection data into maintenance workflows. The University of Cincinnati assisted with sensor integration, led development of several damage-detection features, and supported system maintenance during the trial period. Together, the public, private, and academic partners evaluated not only detection accuracy but also how the information could be operationalized by a state transportation agency.

As the test vehicles collected data, potential roadway deficiencies were identified using edge-based AI models running on the vehicles themselves. That data was then transmitted to a Honda-managed cloud platform, where additional processing and analysis took place. From there, the information was integrated into Parsons’ asset management system, which allowed ODOT staff to view findings through web-based dashboards. The system was designed to automatically generate maintenance work orders that could be prioritized by severity and grouped geographically to improve efficiency.

ODOT compared the automated detections against its existing visual inspection processes. According to the results released by the project team, the system demonstrated high accuracy in several key categories. Damaged or obstructed road signs were identified with an accuracy rate of 99 percent. Detection of damaged guardrails reached 93 percent accuracy, while pothole detection averaged 89 percent across the road types included in the study. Shoulder drop-offs, which can be difficult to identify consistently through manual inspection, were also flagged with a high degree of reliability.

In addition to identifying discrete defects, the system was able to assess broader roadway conditions. Vehicle sensor data was used to measure road roughness levels, providing quantitative information that could support pavement management decisions. Lane marking quality was also evaluated, with results suggesting that only a small percentage of the surveyed roads had insufficient striping. Project leaders indicated that this type of data could allow transportation agencies to refine restriping schedules rather than relying on fixed intervals or subjective assessments.

A key component of the pilot was the development of a feedback loop to improve system performance over time. ODOT staff were able to flag false positives or missed detections within the dashboard interface. Those inputs were then fed back into the AI models to refine detection accuracy. This approach reflects a broader trend in infrastructure monitoring toward adaptive systems that improve as they are exposed to more data and user feedback.

Beyond detection performance, the project evaluated potential operational and financial impacts. By reducing the need for manual roadway inspections, the system could limit the amount of time maintenance workers spend exposed to live traffic, addressing a long-standing safety concern for transportation agencies. Fewer in-person inspections also translate into lower labor costs and reduced use of specialized inspection vehicles.

Based on the pilot results, the project team estimated that automated road condition detection could save ODOT more than $4.5 million annually. Those projected savings are attributed to reduced inspection labor, more efficient maintenance scheduling, and the ability to address issues earlier, potentially avoiding more expensive repairs caused by deferred maintenance. While the estimate is specific to Ohio’s roadway network and operating costs, it is presented as an example of how similar systems could offer financial benefits if scaled to other jurisdictions.

The pilot also explored how vehicle-generated data could fit into existing asset management practices rather than replacing them entirely. Project partners emphasized that production vehicle sensors are primarily designed for driving and safety systems, not infrastructure monitoring. As a result, specialized analytics are required to account for sensor placement, calibration differences, and the variability inherent in crowd-sourced data. The Ohio pilot was intended to demonstrate how these limitations can be addressed through tailored data processing and integration with established inspection programs.

Looking ahead, Honda and its partners are examining options to scale the Proactive Roadway Maintenance System beyond a limited pilot. One concept under consideration involves anonymized data sharing from customer-owned vehicles equipped with compatible sensors. In that model, everyday driving could contribute to a continuously updated picture of roadway conditions, supplementing agency-operated inspection fleets. No deployment timeline or implementation details have been announced, and any such approach would depend on regulatory, privacy, and technical considerations.

From Honda’s perspective, the project aligns with its broader safety initiatives and research into connected and automated vehicle technologies. The company has stated that improving roadway conditions can complement in-vehicle safety systems by reducing hazards that contribute to crashes or degrade the performance of driver-assistance features such as lane-keeping systems. Faded lane markings, for example, can limit the effectiveness of camera-based assistance technologies, making infrastructure quality a relevant factor in vehicle safety performance.

For DriveOhio and ODOT, the pilot fits within ongoing efforts to evaluate smart mobility technologies that could modernize transportation operations. State transportation agencies across the U.S. are facing growing maintenance demands, constrained budgets, and workforce challenges. Automated inspection tools are increasingly being studied as a way to extend limited resources while maintaining or improving safety outcomes.

The Ohio pilot did not result in an immediate statewide deployment, but it provided a large-scale data set and operational insights that ODOT can use to inform future decisions. Project leaders characterized the effort as a proof of concept that demonstrated technical feasibility and measurable benefits under real-world conditions, rather than a finalized production system.

The collaboration between an automotive manufacturer, a state transportation agency, a university, and multiple technology providers was described by participants as a central factor in the project’s success. Each partner contributed domain-specific expertise, from vehicle sensor integration and AI development to infrastructure engineering and asset management workflows. Similar cross-sector partnerships are likely to play a role in future transportation technology pilots, particularly those that sit at the intersection of vehicles and public infrastructure.

As transportation agencies continue to explore data-driven maintenance strategies, the Ohio pilot offers one example of how vehicle-based sensing could be incorporated into roadway management. While further testing and refinement would be required before large-scale adoption, the results suggest that automated detection systems can achieve a level of accuracy sufficient to support operational decision-making when combined with existing inspection practices.