Research Question
Can Public Geometry Support Prospective Comparison?
R.I.S.K. tests whether road shape, physics-based reasoning, and controlled assumptions can identify relative differences before collision-history data is introduced.
Prospective Relative-Risk Model
Roadway Infrastructure Safety Kinematics
Road Risk converts public road geometry into an inspectable physics-based model. Select a road, derive its geometry, apply scenario assumptions, and compare the resulting model output against wider road-network samples.
Model boundary
Model outputs are comparative estimates based on public road geometry, physics-informed calculations, and selected assumptions. They are not official crash predictions or road-safety ratings.
Annualised Comparative Model Output
--
Generated in the live app after a road is selected and scenario assumptions are set.
Route Analysis
Distance, duration, sampled risk, hotspots, route overlays, and exports live inside the app.
Illustrative preview, not a reviewed case study. Static overview of the live application outputs: selected-road geometry, comparative model output, route context, and exportable evidence.
Project at a Glance
What Should Be Understood First
Data Source
OpenStreetMap-Derived Road Geometry
The live app uses public OSM-derived ways, nodes, tags, and selected-segment geometry through the public mapping data pipeline.
Core Physics
Curvature, Radius, Safe Speed, and Stopping Distance
Coordinate geometry is converted into kinematic outputs that describe turning demand and stopping-distance sensitivity.
Scenario Testing
Rain, Fog, Fatigue, Overspeed, Vehicle Type, and Surface
The same road can be retested under explicit assumptions to show how model output changes.
Comparative Output
Relative Indicator, Not Crash Prediction
Outputs show whether a road appears typical, elevated, or unusually demanding within a sampled model context.
Limitation
Transparent Prototype Boundary
The model is not an official safety rating and has not been fully calibrated against national collision-history datasets.
For Judges
What to Inspect First
The intended competition route is deliberately short: understand the research question, check the method, inspect saved results, then test the live model.
1 / Project
Research Question and Hypothesis
Start with why the project tests public road geometry and physics-informed assumptions before collision-history data is introduced.
2 / Live Model
Outputs and Inspectability
Open the app, select a road, read the Annualised Comparative Model Output, safe speed, stopping distance, and data-confidence notes.
3 / Results
Evidence Boundary
Use the dashboard to inspect saved model-output records while keeping the boundary clear: not observed crash rates, not official ratings.
Data Pipeline
From Public Road Data to Relative-Risk Output
The public site explains the same pipeline used by the live app. The goal is transparency: what is measured, what is derived, what is assumed, and what the output can responsibly mean.
01
Public Road Data
Road Risk queries public map data and road geometry rather than relying on static screenshots or decorative map tiles.
02
Selected Geometry
The clicked road is projected onto its road polyline so length, heading, radius, and curvature can be derived from geometry.
03
Physics-Informed Model
Curvature, speed, friction, stopping distance, and safe-speed checks are interpreted through transparent formula blocks.
04
Scenario Assumptions
Weather, lighting, vehicle profile, behaviour, traffic exposure, and missing data are handled as explicit assumptions.
05
Relative-Risk Output
The result is a modelled comparison supported by graphs, maths, exports, and clear limits.
Using Road Risk
Responsible Interpretation Workflow
Launch the Application
Open the live map and choose a location.
Select a Road
Click a visible road segment and check the highlighted geometry.
Read Model Outputs
Start with annual output, safe speed, and stopping distance.
Change Scenario Assumptions
Adjust weather, vehicle, speed, visibility, and driver context.
Open Mathematical Detail
Inspect how the current values are derived.
Compare Distributions
Use percentile context before interpreting a raw model output.
Export Results
Preserve geometry, assumptions, values, and evidence.
System Scale
A Research Project Built Around a Live Analytical Application
Source Code
~19.8k
Approximate lines across the live app and public project files.
JavaScript
~15k
Approximate live-app logic for map interaction, modelling, routes, graphs, and exports.
CSS
~4k
Approximate interface styling across the app and public research pages.
Tracked Versions
120+
Iteration history used to develop, test, and refine the prototype.
Development Logs
40
Booklet-style development entries documenting the research and build sequence.
Interface Elements
249
Unique visible controls, panels, outputs, and interface elements across the live application.
Interactive Controls
100+
Scenario, route, graph, export, and interpretation controls exposed to the user.
Vehicle Profiles
12
Vehicle presets used to vary model assumptions and interpretation context.
OSM Attributes
22
Road tags considered across geometry, context, surface, speed, and infrastructure assumptions.
Risk-Factor Defaults
25
Default modelling values used to keep assumptions explicit and reviewable.
Assumption Profiles
3
Scenario profiles for comparing how conditions alter the same road geometry.
Graph Views
9
Distribution views for percentile context, tail behaviour, and comparative interpretation.
Statistical Indicators
18
Summary indicators used to interpret selected roads, distributions, and route-level outputs.
Exports
5
CSV, GeoJSON, JSON, distribution data, and PNG-style reporting outputs.
Output Interpretation
Model Outputs as Part of the Evidence Trail
The live app surfaces several values because no single number should carry the full interpretation.
Annualised Comparative Model Output
Primary Comparative Indicator
Scenario-adjusted model output for the selected road. It appears in the risk card, graphs, mathematical detail, and exports; it is not an observed crash rate or official assessment.
Daily Interpretive View
Time-Scale Interpretation
Converts the annual output into a daily-style view for interpretation. It does not turn the model into an observed daily crash rate.
Safe Speed
Physics-Informed Curve Check
Shows a simplified friction-limited speed estimate for the selected geometry and scenario assumptions.
Stopping Distance
Reaction and Braking Demand
Shows how speed, reaction time, and friction assumptions affect the distance needed to stop.
Infrastructure Rating
Infrastructure Context Indicator
Summarises visible infrastructure/context clues and fallback assumptions. It is not an official road-standard rating.
Percentile Context
Sampled-Network Comparison
Places the selected road relative to sampled roads under the current assumptions so raw model outputs are easier to interpret.
Route Mean / Peak
Route-Level Summary
Route analysis can show mean model output and the highest sampled segment, helping identify local hotspots along a route.
Model Preview
Model Outputs Depend on Visible Assumptions
Road Risk is designed as an inspectable modelling surface. It shows the result, but also the geometry, formula chain, scenario choices, confidence notes, and exportable evidence trail behind it.
Interpretation note
The app does not claim to estimate an observed crash rate for an individual road. It provides a comparative model output that can support research, discussion, education, and early screening.
Primary Output
Annualised Comparative Model Output
A scenario-adjusted comparative value used consistently across the risk panel, maths, graphs, and exports.
Physical Output
Safe Speed and Stopping Distance
Curvature, friction, reaction time, and speed assumptions are translated into interpretable vehicle-motion checks.
Context Output
Percentile and Hotspot Context
A selected road can be compared against sampled roads nearby, making the output less isolated and more interpretable.
Mathematical Basis
Geometry and Vehicle-Dynamics Relationships
These compact examples mirror the physics language used on the Maths page and in the live app's method panels.
Radius
Road Bend Geometry
\[ r = \frac{L}{\theta} \]
Segment length and heading change produce a curve-radius estimate.
Safe Speed
Friction-Limited Speed
\[ v_{\text{safe}} = \sqrt{\mu g r} \]
Friction and radius create a simplified turning-speed check.
Stopping Distance
Reaction and Braking Demand
\[ d_{\text{stop}} = v t_r + \frac{v^2}{2 \mu g} \]
Speed, reaction time, and friction assumptions shape stopping demand.
Applications and Scope
Built for judges, teachers, students, researchers, and safety organisations
Education
Connect Road Geometry to Physics
Students can see curvature, speed, friction, stopping distance, and uncertainty on real roads rather than abstract textbook diagrams.
Research
Inspect a Transparent Pipeline
The model separates measured geometry, public tags, derived quantities, assumptions, and comparative outputs.
Road Safety
Support Early-Stage Screening
Proactive modelling can highlight roads worth inspecting while remaining clear that formal audits require more evidence.
Public Sector
Export and Communicate Evidence
CSV, GeoJSON, JSON, graph, and map outputs make the analysis easier to review, discuss, and reproduce.
Site Structure
Review the Method, Then Test the Live Model
Overview and Judge Guide
Research question, hypothesis, what was built, responsible use, timeline, and judge workflow.
Data, Formulas, Assumptions, and Graphs
One consolidated guide to the model pipeline, scenario controls, data quality, and validation boundary.
Results Dashboard
Saved cases, graphs, evidence-quality notes, scenario comparisons, and clear interpretation boundaries.
Evidence Base
Booklet-aligned sources, technical documentation, licensing, and attribution.
Test a Road Segment
Select a road, change one assumption, inspect the output, and save or export a model-output case.
Live Application
Open the Road Risk Live Model
Select a road, inspect the model output, change assumptions, generate graphs, analyse routes, and export the same evidence trail described across this site.