Peachtree Corners’ newly announced Autonomous Mobility Program, offering up to $250,000 to companies validating autonomous mobility technologies on public roads, reflects a broader shift in urban transportation: cities are no longer treating mobility systems as static infrastructure. They are becoming active operating environments where shuttles, delivery vehicles, connected intersections, pedestrians, retail trips, and commuter traffic interact in real time.
That kind of mixed urban setting exposes a practical engineering problem. Before a city adds new mobility services, adjusts curb access, retimes signals, or considers geometric improvements, it needs to know where the road network is actually failing. Bottleneck identification is not simply a matter of finding the slowest link on a map. It requires separating demand problems from management problems, recurring congestion from occasional disruptions, and isolated delay from network-level time loss.
Ticon’s approach starts with the premise that urban bottlenecks are spatiotemporal, not just spatial. A segment may operate acceptably for 20 hours a day and become the controlling constraint for the network during a 45-minute school, commuter, retail, or event peak. Another corridor may show moderate speeds but accumulate large total driver time loss because volume is high throughout the day. Ranking these locations correctly requires volume, speed, saturation, and delay to be measured together.
Ticon’s platform builds that view from broad coverage and fine resolution. According to the Ticon Methodology documentation, the system provides almost 100% road network coverage, including more than 97% of roads at Functional Road Class 6 and above, and 100% time coverage. It consolidates permanent and portable detector data, traffic counters, GPS data, connected vehicle data, GIS, demographics, traffic organization information, event transformations, and other sources. After cross verification, filtering, and processing through Ticon’s proprietary algorithms, the platform estimates speeds, volumes, and related derivatives for 95% of roadways, with road segment resolution as short as 35 feet and about 225 feet on average. Time resolution can reach 5-minute intervals, and in many cases 15 seconds.
That resolution matters because bottlenecks are often missed by coarse methods. A corridor average can hide a failing intersection approach. A 48-hour count can miss weekly variability. A planning model can show demand growth but fail to reveal whether the delay is caused by signal timing, turning demand, access friction, or saturation on a short connector link. Ticon’s methodology combines dynamic speed-density relationships, machine learning modeling, saturation flow analysis, automated impact analysis, and multi-source traffic data consolidation to move from “traffic is slow here” to “this specific link, approach, or period is producing measurable delay.”
The distinction is important for intervention planning. Ticon’s mobility research, described in Traffic congestion, what works, what doesn’t, analyzed traffic flows at 126 road intersections across nine U.S. states using about 200 million data points. The study used the COVID-era reduction in traffic demand as a natural experiment, observing traffic demand decreases of up to 30% or more. Yet delay on signalized roads fell by much less than volume. In some cases, traffic delay changed little even when demand was reduced almost by half. The engineering implication is direct: not every bottleneck is solved by reducing demand, and not every congested location requires expensive widening. Poor traffic organization can preserve delay even when volumes fall.
For city engineers, that finding changes how bottleneck screening should be performed. A traditional capacity-first workflow tends to ask whether a road needs more lanes or whether a corridor needs a larger capital project. An evidence-based workflow asks a sequence of more precise questions: Where is delay generated? During which intervals? Is the controlling factor volume, saturation, signal operation, intersection geometry, turning distribution, access management, or a network interaction upstream or downstream? Which location contributes most to total driver time loss across the area?
Ticon’s virtual transportation model is designed to support that type of ranking. The model can generate numerical comparisons of road sections using metrics such as traffic delay, saturation degree, total driver time loss, and the impact of each section on area mobility. For each segment and each time interval, the platform can provide parameters such as AADT, intraday speed, and intraday volume profiles. In practice, this means a city can compare a congested arterial, a failing left-turn approach, and a short downtown connector on common performance measures rather than relying on complaints, isolated counts, or windshield observations.
Volume accuracy is central to this process because bottleneck severity cannot be assessed from speed alone. A low-speed, low-volume street may inconvenience a few drivers. A moderate-speed, high-volume arterial may impose thousands of vehicle-hours of delay across a week. In Brodski and Chaihorsky’s AADT Estimation by Various Methods: Accuracy and Reliability (2018), Ticon’s AADT estimation was evaluated across Georgia, Nevada, and California. The study selected 695 counting points, rejected 28 due to detector-counting errors and 30 due to gaps in GPS data, and used the remaining 637 points, most of them bidirectional, for more than 1,200 estimations. The reported AADT performance included a median average percentage error of 4.78% and a relative root mean square error of 11.97%, keeping expected AADT error within 20% boundaries with 90% confidence.
More recent field verification also supports the use of Ticon estimates for operational screening. In Field Verification of Ticon Traffic Flow Volume Estimation Accuracy, Ticon compared its estimates with independent detector data from State Departments of Transportation. The study reported AADT average error of 8.31% and AADT estimation error of 19.40% with 90% confidence. For hourly volumes, average error was 10.1%, with hourly volume estimation error of 19.94% at 85% confidence. For bottleneck identification, this level of accuracy is particularly useful because the task is often comparative: identify which links, approaches, or time periods are producing the largest mobility losses and therefore deserve detailed engineering attention.
Turning movements add another layer. Many urban bottlenecks are intersection problems before they are corridor problems. A through lane may appear under capacity while a left-turn movement blocks progression. A right-turn movement may conflict with pedestrian activity. An autonomous shuttle route may be feasible in one direction but sensitive to a peak-period turning surge at a mixed-use entrance. Ticon’s Ticon Turns: Verification of Accuracy describes a proprietary turning movement estimation method based on multivariate analysis of GIS, traffic events and management data, demographics, connected vehicles, traffic organization, location-based services, traffic detection, GPS, navigation data, and other inputs. The platform estimates right, left, and through movement demand for each 15-minute period across a full 24-hour day, with aggregation available by weekday, weekend, month, season, year, peak period, or off-peak period.
In the Maine verification study cited in that report, Ticon estimates were compared with portable detector measurements from AVCOG at two three-leg and two four-leg intersections. The discrepancy between Ticon estimates and detector measurements was within a 7% to 22% range. The report notes that short-term measurements themselves can vary by roughly plus or minus 25%, which is why extended observation is important when deciding where interventions are needed. For cities, the practical benefit is the ability to screen turning demand over longer periods before sending crews, consultants, or hardware to the field.
This matters for programs like autonomous mobility deployment, but it matters just as much for conventional urban operations. Adding a shuttle, modifying a curb lane, opening a new retail anchor, changing signal timing, or shifting truck routes can all redistribute demand at intersections. Without a baseline picture of segment-level and movement-level performance, an intervention intended to improve mobility in one location can move the bottleneck one block downstream.
Ticon’s congestion workflow also helps distinguish between intervention types. If a link shows recurring speed collapse, high volume, and high saturation during a narrow peak, signal timing or coordination may be the first candidate. Ticon’s congestion research reports that up to 50% reduction in travel delay is achievable through signal timing optimization alone, based on Ticon’s project experience. If a location shows high total driver time loss throughout the day, a larger operational or geometric review may be warranted. If a corridor exhibits delay despite lower demand, the issue may be traffic organization rather than capacity. If a new mobility service is planned, the same metrics can identify where the service is likely to be delayed, where it may interfere with existing traffic, and where priority treatments would have the greatest network benefit.
The financial implications are also clear. Congestion imposed more than $87 billion in lost productivity costs in the United States alone in 2019, according to Ticon’s congestion analysis. Yet technology deployment remains limited relative to the scale of the network. The same Ticon document notes that only about 150 adaptive signal control technology installations were completed at intersections in the United States over the previous 10 years, compared with roughly 355,000 signalized intersections nationwide. One reason is uncertainty: agencies often do not know which locations will benefit most from a given ITS investment. Bottleneck ranking by delay, saturation, volume, and areal mobility impact helps reduce that uncertainty.
There are also limits that responsible analytics must acknowledge. Ticon’s platform can compute many of the metrics used in practice and recommended by the Highway Capacity Manual, including delay, travel time delay, saturation degree, and speed-volume relationships. However, the current data model does not directly determine some parameters, such as the number of stops or queue length. That does not reduce the value of network-level screening. It clarifies the workflow: use Ticon to identify and rank likely bottlenecks, then apply field validation, simulation, or design analysis where the ranking shows the greatest expected return.
For urban networks, the strongest intervention strategy is rarely “fix the worst-looking intersection.” It is to identify the locations where small operational changes can produce the largest reduction in total delay, where recurring saturation threatens network reliability, and where future mobility services may intensify existing constraints. As cities test autonomous shuttles, expand mixed-use districts, and manage growing curb and intersection demands, bottleneck identification becomes a continuous engineering function rather than a one-time study.
The practical goal is not just to see congestion. It is to measure where congestion is created, quantify how much driver time it consumes, understand when and why it appears, and select interventions in the order that produces the greatest mobility benefit. That is where Ticon’s high-resolution traffic analytics, validated volume estimation, turning movement modeling, and saturation-based screening give planners and engineers a more precise foundation for urban road network improvement.
Join the network to access Ticon case studies and to receive Ticon newsletters
@2023 Ticon
