County road commissions, municipal street departments, townships, and local road agencies across Michigan and the broader Midwest rely on RoadSoft as the operational core of their local road and asset management programmes. Developed by the Michigan Technological University Transportation Institute (MTU-T2), RoadSoft is purpose-built for the scale, workflow, and reporting requirements of local agencies the county engineers, road commission directors, and public works superintendents who manage thousands of lane miles with lean staffing, tight budgets, and direct accountability to local elected officials and communities.
But one question is coming up more frequently across county road commissions and municipal street departments: how do we bring into RoadSoft without disrupting the PASER rating workflows and local asset management practices our teams have built over years?
This guide is written specifically for county engineers, road commission managers, local agency asset managers, and GIS staff who are already working within the RoadSoft ecosystem and are evaluating how AI-powered road condition assessment fits into not on top of their existing local road management setup.
.webp)
Before covering where AI fits, it is worth being precise about what RoadSoft actually provides for local road management because this determines exactly where an AI condition data layer connects.
RoadSoft is a local agency road and asset management platform developed specifically for county road commissions, municipalities, and townships managing local road networks. Unlike state DOT-scale platforms designed for large bureaucracies, RoadSoft is built around the operational realities of local agencies: small engineering teams, limited GIS capacity, PASER-based condition rating workflows, Act 51 reporting requirements in Michigan, and the need to produce defensible capital programmes and budget justifications for local governing boards on modest data infrastructure.
The foundation of RoadSoft’s pavement management capability is the PASER (Pavement Surface Evaluation and Rating) system the 10-point visual condition rating scale developed by the University of Wisconsin Transportation Information Center and adopted as the standard condition assessment methodology across Michigan and much of the Midwest. RoadSoft is designed to store, manage, and analyse PASER ratings at the road segment level, tracking condition trends over time, projecting deterioration under different maintenance scenarios, and identifying the segments where maintenance investment will deliver the greatest network benefit.
RoadSoft maintains inventories of road segments, bridges, culverts, signs, pavement markings, drainage structures, and other roadway assets at the local network level. Each asset record carries location, classification, condition, and maintenance history. For county road commissions managing networks of several hundred to several thousand miles, RoadSoft’s asset inventory is the authoritative record of what the agency owns, where it is, and what condition it is in.
Capital Improvement Planning and Budget Analysis
RoadSoft’s capital improvement planning tools enable local agencies to develop multi-year road improvement programmes based on pavement condition data, deterioration projections, treatment costs, and available budget. The platform can model the network condition outcome under different budget scenarios showing local boards and elected officials what the road network will look like in five or ten years if funding is maintained, reduced, or increased. This scenario modelling is what local agency engineers use to justify capital budget requests and communicate infrastructure needs to non-technical audiences.
In Michigan, local road agencies receiving Act 51 transportation funding are required to submit annual road condition data and asset reports to the Transportation Asset Management Council (TAMC). RoadSoft is the platform most widely used by Michigan county road commissions and municipalities to compile and submit these TAMC reports which include PASER-based condition ratings for the local road network, asset inventory data, and maintenance activity records. Compliance with Act 51 reporting requirements is a legal obligation for agencies receiving state transportation funding, making RoadSoft’s TAMC reporting workflow a non-negotiable operational function.
RoadSoft integrates with GIS platforms, including Esri ArcGIS, to display road condition data, asset locations, and maintenance records spatially. The RoadSoft GIS module allows agencies to view condition ratings, work history, and asset data on an interactive map enabling visual identification of problem areas, maintenance zone planning, and spatial reporting for board presentations and public communications. Pavement condition heatmaps generated from RoadSoft PASER data are commonly used in local budget and capital programme presentations.
RoadSoft tracks maintenance activities and work orders against road segments and asset records, providing a history of what treatments have been applied to each road section and when. This maintenance history feeds into deterioration modelling and treatment effectiveness analysis allowing agencies to understand which treatments are delivering expected performance on their specific road network under local traffic and climate conditions.
RoadSoft includes a dedicated sign management module that tracks the location, type, condition, retroreflectivity status, and replacement history of every sign in the agency’s inventory. Federal retroreflectivity compliance requirements under FHWA’s Maintaining Traffic Sign Retroreflectivity rule mean that sign inventory management is a compliance obligation for local agencies, not just an operational convenience. RoadSoft’s sign module provides the structured record required to demonstrate and document compliance.
RoadSoft manages local road and asset data with a precision and workflow fit that no other platform matches for local agencies. What it does not do is generate that data.
The PASER ratings, asset condition records, and maintenance histories in RoadSoft are only as current as the last time agency staff or a contractor physically drove the network and recorded the results. In practice, for most county road commissions and municipal street departments, that means:
Most local agencies conduct full-network PASER surveys on a two- to four-year cycle or less frequently on lower-volume roads. Between survey cycles, condition records in RoadSoft remain static. A road that has deteriorated significantly since the last survey still carries its old PASER rating in the system, meaning maintenance decisions and capital programme prioritisation are based on condition data that may be years out of date.
The PASER system relies on trained raters making visual assessments from a moving vehicle. Two raters assessing the same road on the same day will frequently assign different PASER scores particularly in the middle of the 1–10 scale where the boundary between ratings is a matter of professional judgement. Year-on-year condition series assembled from different raters, or the same rater in different seasons, contain noise that makes it difficult to distinguish genuine deterioration from rating variability.
Local agencies operate with small engineering teams. Conducting a full-network PASER survey is a significant staff time commitment that competes with the day-to-day demands of road maintenance operations. Many agencies struggle to achieve even biennial full-network coverage, let alone the annual or more frequent surveys that would make deterioration modelling and preventive maintenance scheduling genuinely data-driven.
Signs are damaged or stolen, markings fade, culverts deteriorate, and drainage structures fail but RoadSoft’s asset records only reflect these changes when a crew member notices and logs the update. For agencies managing thousands of signs and hundreds of miles of markings across a large county road network, maintaining a current inventory record through manual observation alone is operationally impractical.
The PASER system is designed for visual assessment of visible surface condition. Early-stage cracking, minor rutting, and subsurface drainage failures the precursors to rapid deterioration are frequently below the threshold of reliable visual detection from a moving vehicle. By the time they show up in PASER ratings, the road has often deteriorated past the point where low-cost preventive treatments crack sealing, chip sealing are still viable. Reconstruction costs replace what preservation costs could have prevented.
Michigan’s TAMC reporting requirements mandate annual condition data submission. For agencies that cannot achieve annual full-network surveys with available staff and budget, TAMC submissions may rely on interpolated, carried-forward, or estimated condition data reducing the accuracy of statewide condition reporting and the credibility of the local agency’s funding case.
The result: a well-structured, operationally active road management platform populated with condition data that may be two to four years old, rated with variable consistency, and missing the early deterioration signals that if caught in time would allow preservation treatments to extend pavement life at a fraction of the cost of the reconstruction that follows.
This is not a criticism of RoadSoft. It is a structural limitation of how local road condition data has historically been collected. The platform can only manage what it receives.
AI-based computer vision for road surveys addresses the condition data generation problem directly and does so in a way that is particularly well-suited to the resource constraints of local road agencies. The approach is operationally simple: a vehicle-mounted AI dashcam survey footage captures geo-tagged, time-stamped video of the road network as the vehicle drives at normal operating speeds. That footage is processed through AI models that detect and classify road conditions and assets automatically, returning structured results that load directly into RoadSoft.
No specialist survey vehicle. No dedicated survey crew. No additional staff time beyond mounting a camera on an existing vehicle. Any vehicle already operating on the road network a road commission maintenance truck, a supervisor’s pickup, an inspection van already driving the county roads daily becomes a survey platform. For local agencies where staff time is the binding constraint on survey frequency, this operational simplicity is the critical difference.
RoadVision AI’s models are trained on over 100 million road images from networks across South Asia, the Middle East, Southeast Asia, Europe, and Africa, covering diverse pavement types, climate conditions, road classifications, and asset configurations. Importantly for Midwest local road agencies, the training data includes gravel and unpaved road surfaces, freeze-thaw damaged pavement, chip seal and microsurface treatments, and low-volume rural road conditions the surface types that dominate many county road networks and that general-purpose AI models trained primarily on urban pavement frequently underperform on.
A single dashcam survey pass, processed through the RoadVision AI pipeline, returns structured condition intelligence across the following categories all mapped to RoadSoft’s data schema:
Detection and classification of surface distress including potholes, longitudinal cracking, transverse cracking, alligator (fatigue) cracking, rutting, edge deterioration, patching quality, ravelling, and surface delamination. Each finding is severity-scored (Low / Medium / High / Critical) and contributes to a per-segment condition score that maps to the PASER 1–10 rating scale the native condition index in RoadSoft. An IRI-equivalent roughness value is also returned per 100-metre segment. For agencies that want both AI-derived PASER equivalents and raw distress data, both are delivered in the same output file.
Automated detection and classification of 80+ road asset types: regulatory signage, warning signs, informational signs, road markings (centreline, edge line, stop bars, pedestrian crossings, school zone markings, chevrons), kerb and edge conditions, guard rails and safety barriers, crash attenuators, drainage structures, culverts, driveway approaches, lighting columns, and mailbox approaches. Every road asset detection includes GPS coordinates, asset type, condition grade, and a photographic evidence frame ready to create new records or update existing records in RoadSoft’s asset inventory and sign management modules.
AI detection of signs includes condition grading assessing physical damage, fading, obstruction, and mounting condition alongside retroreflectivity status assessment based on visual analysis of sign face brightness relative to expected values. For local agencies managing FHWA retroreflectivity compliance obligations through RoadSoft’s sign module, AI-derived sign condition and retroreflectivity assessments provide a systematic, network-wide compliance check that manual inspection of individual signs cannot achieve efficiently at scale.
Detection of safety deficiencies including faded or absent pavement markings, damaged or missing safety barriers, vegetation encroachment onto carriageway clearances, obstructed sight lines at intersections, and right-of-way intrusions. Each safety finding is geo-tagged and severity-scored, providing the documentation record that local agencies need to demonstrate awareness of known safety deficiencies and timely response.
Detection of visible drainage deficiencies blocked culvert inlets and outlets, failed ditch grades, eroded shoulders adjacent to drainage structures, and standing water indicators alongside culvert location confirmation and condition grading. For county road agencies where drainage maintenance is one of the highest-volume maintenance activities, AI detection of drainage deficiencies provides a systematic basis for prioritising drainage work orders in RoadSoft.
Every detection record includes: defect or asset type, severity or condition grade, confidence score, GPS coordinates, route reference, timestamp, and a geo-tagged photographic evidence frame formatted for direct import into RoadSoft.
RoadSoft manages road condition and asset data through structured segment records, PASER rating histories, asset inventory tables, and work order records with data import interfaces designed for local agency workflows. RoadVision AI outputs are structured to map directly to these schemas, enabling AI-generated road condition data to enter RoadSoft’s management and reporting workflows without manual re-entry or data transformation.
This is the most critical alignment point for RoadSoft users. RoadVision AI processes detected distress types, severity levels, and surface condition indicators and maps the result to an AI-derived PASER equivalent score on the standard 1–10 scale, at the road segment length defined in RoadSoft. This means AI survey outputs load into RoadSoft as PASER ratings the native condition unit feeding directly into deterioration curves, capital improvement planning, and TAMC reporting workflows without requiring any condition index conversion or recalibration. Agencies that want to cross-reference AI-derived PASER scores against manual PASER ratings from the same survey cycle can do so directly, using the same scale.
AI condition results are delivered at the segment level, referenced by the road segment identifiers and from/to references already in the agency’s RoadSoft database. This means each AI survey result maps to an existing RoadSoft road record without requiring manual spatial matching, GIS processing, or database administrator intervention. Condition scores, distress flags, and survey dates load against the correct segment records automatically.
For new assets detected by the AI survey that are not yet in RoadSoft’s inventory, detection outputs include all attributes needed to create a new record in the RoadSoft asset inventory or sign management module: asset type, GPS location, condition grade, detection date, and evidence photo. For existing asset records, AI condition outputs update the current condition field and append a new dated assessment record maintaining the full condition history that RoadSoft’s maintenance planning and FHWA compliance tracking depend on.
AI survey outputs are structured to align with the condition data inputs required for Michigan TAMC annual reporting. PASER ratings by road segment, survey date, road classification (primary, local, urban), and functional class are delivered in the format that RoadSoft compiles for TAMC submission. For agencies currently unable to achieve annual full-network manual surveys, AI surveys provide the annual condition data that TAMC reporting requires without requiring additional staff time or contracted survey costs at the scale of traditional manual campaigns.
AI survey results are available as Shapefile and GeoJSON exports compatible with RoadSoft’s GIS module and Esri ArcGIS integration. Pavement condition scores, defect locations, and asset detection points load into the GIS layer as spatially referenced features — enabling condition heatmaps, defect density maps, and asset inventory maps to be generated directly within the RoadSoft GIS environment. These spatial outputs are exactly the board presentation materials that county engineers use to communicate road network condition to local commissioners and elected officials.
Once loaded, AI road survey data participates in every downstream RoadSoft workflow the agency already runs: pavement condition trend analysis, deterioration modelling, capital improvement programme generation, work order prioritisation, TAMC report compilation, sign compliance tracking, and GIS-based board presentations. Agency staff work in the same RoadSoft environment they have always used the difference is that condition data is current, consistent, and covers the full network every survey cycle.
The evaluation path is designed to be practical for local agency teams with limited IT and GIS capacity:
For local road agencies evaluating AI survey tools, the practical question is not whether AI detection is technically viable — it demonstrably is. The question is whether it works on the road types the agency actually manages, produces condition scores in the format the agency’s platform actually uses, and fits into the operational reality of a lean local agency team. That is what this integration is specifically designed to address.
What does not change: The RoadSoft platform. The PASER rating system. The agency’s data structure. The TAMC reporting workflow. The capital improvement planning process. The sign management module. The GIS integration. The board presentation formats.
What changes: The frequency, coverage, consistency, and staffing cost of the condition data feeding into those workflows.
Survey frequency increases without additional staff burden. Because any agency vehicle already driving the road network can serve as a survey platform, PASER surveys no longer compete with other staff time demands. A county road commission currently conducting biennial surveys can move to annual or more frequent surveys without additional engineering staff. Roads that are currently surveyed on a three- or four-year cycle because of staffing constraints can be assessed annually as a matter of routine fleet operations.
Coverage becomes complete rather than prioritised. Local agencies with limited survey capacity typically prioritise primary roads and known problem areas. AI processing of fleet dashcam footage covers every road in the agency’s jurisdiction on every survey pass including low-volume township roads, subdivision streets, and dead-end rural sections that have not received a formal PASER assessment in years. RoadSoft condition records that have been static for extended periods receive current data.
PASER ratings become consistent and defensible. One of the practical limitations of manual PASER surveys is that condition ratings reflect the individual rater’s judgement on that day. When the same road is rated differently in consecutive years by different staff, it is difficult to distinguish genuine deterioration from rater variability. AI-derived PASER equivalents are produced by the same models applying the same criteria on every survey run making year-on-year condition comparisons genuinely meaningful and deterioration trends statistically reliable.
TAMC reporting becomes data-driven rather than estimated. With AI surveys providing annual full-network condition data, Michigan agencies can submit TAMC reports based on current measured condition rather than interpolated or carried-forward ratings from previous survey cycles. This improves both the accuracy of statewide condition reporting and the credibility of the local agency’s condition data in the context of Act 51 funding distribution.
Preservation treatments reach roads before reconstruction becomes the only option. Frequent AI surveys detect early-stage cracking, surface distress, and drainage deterioration at PASER ratings of 6 or 7 when crack sealing, chip sealing, and surface treatment are still viable and cost-effective. RoadSoft’s capital improvement planning tools can then prioritise these roads for preservation treatment before they deteriorate to PASER 4 or below, where the treatment options are overlay or reconstruction at substantially higher cost.
Sign compliance management becomes systematic. Rather than relying on staff to notice and report sign damage and fading during routine operations, AI surveys provide a systematic, network-wide sign condition and retroreflectivity assessment on every survey pass. RoadSoft’s sign module receives updated condition records, and FHWA retroreflectivity compliance tracking reflects current assessed status rather than last-inspection dates.
Any GPS-enabled dashcam producing standard MP4 or MOV video at 1080p or above. No proprietary hardware, no vehicle modifications beyond mounting the camera. A dashcam that costs under $200 and attaches to the windshield of any road commission vehicle is sufficient. No calibration rig or specialist mounting equipment is required.
No. Any vehicle already operating on the road network serves as a survey platform. A road commission superintendent’s pickup truck, a maintenance crew vehicle, or a supervisor’s van driving regular routes can capture survey footage as part of normal operations. No additional survey scheduling, dedicated survey crew time, or specialist vehicle equipment is required.
RoadVision AI processes detected distress types and severities through a PASER conversion model that maps AI-measured surface condition to the standard PASER 1–10 scale. The conversion is calibrated against ground-truth PASER assessments on Michigan and Midwest road networks. AI-derived PASER equivalents load into RoadSoft’s condition rating fields as standard PASER scores — feeding deterioration models, capital improvement planning, and TAMC reporting without any condition index conversion on the agency side. Agencies wanting to validate AI-derived PASER scores against manual ratings can run parallel assessments on a reference section during the evaluation period.
Yes. AI survey outputs are formatted to align with the condition data inputs required for Michigan TAMC annual submissions in RoadSoft — including PASER ratings by segment, survey date, road classification, and functional class. For agencies that currently struggle to achieve annual full-network manual surveys, AI surveys provide the annually updated condition data that TAMC reporting requires without proportional increases in survey staff time or contracted survey costs.
Standard processing turnaround is 24–48 hours from footage submission, depending on survey volume. For county road commissions surveying their full network in a single campaign, the processed dataset is returned within two to three business days and is ready for immediate import into RoadSoft.
The AI models operate at over 97% detection accuracy across core pavement distress and asset categories on standard road network footage. On validation tests against concurrent manual PASER ratings on Michigan and Midwest road networks, AI-derived PASER equivalents fall within one PASER point of manual ratings in the large majority of cases. Accuracy detail by distress type, surface category, and camera placement configuration is available in the technical documentation provided at onboarding.
The Autonomous Road Engineer
AI that is transforming road infrastructure, from reactive to predictive.
.webp)