Public Data
Open Road Geometry
The project starts with inspectable public road data rather than hidden proprietary inputs.
Development Overview
A Live Research Tool for Road-Risk Modelling
Road Risk combines this public research site with a live app that lets users test selected roads, change assumptions, inspect graphs, and export results. This page explains the build, data sources, method, limits, and next development steps.
Competition Summary
The Project at a Glance
- Title
- R.I.S.K. - Roadway Infrastructure Safety Kinematics.
- Question
- Can public road geometry, physics-informed calculations, and controlled assumptions support prospective relative-risk comparison?
- Hypothesis
- Geometry plus kinematic reasoning can produce useful comparative signals before collision-history calibration is introduced.
- Built
- A public research site and live analytical app for selected-road modelling, scenario testing, graphing, saving, and exporting case records.
- Tested
- Internal model behaviour across curvature, radius, friction, speed, stopping distance, assumptions, and comparative statistics.
- Boundary
- Outputs are comparative model indicators, not official safety ratings or crash predictions.
Project Overview
A Prospective Relative-Risk Investigation
R.I.S.K. stands for Roadway Infrastructure Safety Kinematics. The project investigates whether public road geometry, physics-informed calculations, and controlled scenario assumptions can support early-stage comparative interpretation before collision-history data is introduced.
Road Risk is the live application built for the investigation. It lets users select road segments, inspect derived geometry and model outputs, change scenario assumptions, compare distributions, analyse routes, and export results for review.
Model boundary
Road Risk produces comparative model outputs from public geometry and selected assumptions. These outputs are not official safety ratings or crash predictions.
Originality
What Makes This Project Distinctive?
Physics Chain
Kinematic Interpretation
Curvature, radius, safe-speed, and stopping-distance relationships are exposed as part of the explanation.
Sensitivity
Scenario-Based Modelling
Rain, fog, fatigue, overspeed, vehicle type, surface, and lighting assumptions can be varied deliberately.
Statistics
Comparative Context
Distribution views and percentiles help prevent one raw model value from being read in isolation.
Live App
Inspectable Research Tool
The app lets judges test a selected road, inspect the method, and export a case record.
Research Question
Can Public Road Form Provide an Early Comparative Signal?
Road Risk asks whether selected road geometry, vehicle-dynamics relationships, and transparent assumptions can identify relative changes in model output before collision-history data is used.
Central Hypothesis
Geometry Plus Physics Can Support Prospective Comparison
The hypothesis is treated cautiously: public geometry and kinematic reasoning can support comparative analysis within a defined model boundary, but stronger operational claims require empirical calibration and professional review.
What Was Built
A Public Research Site and a Live Modelling Application
Public Site
Research Site
Explains project scope, method, assumptions, results framing, references, and responsible interpretation.
Live App
Interactive Road Selection
Uses a map interface to select roads, derive geometry, apply scenario assumptions, display outputs, and export evidence.
Method Hub
Technical Explanation
Combines data pipeline, formulas, assumptions, graph interpretation, reliability, and key terms into one navigable method page.
Case Evidence
Reviewed Static Dataset
The app can save opt-in cases locally for browser review and export case-evidence JSON. Only checked and committed cases become part of the reviewed public dataset.
System Scale
System Scale and Auditability
These figures are approximate public-project indicators rather than claims of scientific validation. They describe the implementation scale and app features.
Source Code
~19,800
Approximate lines across the current source files.
JavaScript
~15,000
Approximate logic for road selection, modelling, route analysis, graphs, and exports.
CSS
~4,000
Approximate styling for the live app, public pages, responsive layouts, and panels.
Tracked Versions
120+
Development history and iteration count as a project-scale signal.
Development Logs
40
Research and build notes to support transparency.
Interface Elements
249
Unique output areas, controls, status panels, and interface components in the prototype.
Interactive Controls
100+
Scenario, model, route, graph, export, and UI controls in the live app.
Vehicle Profiles
12
Vehicle presets that change model assumptions rather than labels alone.
OSM Attributes
22
Referenced road attributes across surface, speed, geometry, context, and infrastructure.
Risk-Factor Defaults
25
Default values used to make baseline assumptions visible before scenario changes.
Assumption Profiles
3
Baseline and stress-style profiles for comparing model sensitivity.
Graph Views
9
Statistical views for distribution and percentile interpretation.
Statistical Indicators
18
Indicators used in the distribution, percentile, route, and summary outputs.
Export Formats
5
CSV, GeoJSON, JSON, distribution data, and image-style outputs.
Development Timeline
From Prototype to Research Tool
Stage 1
Project Definition and Core Variables
Defined the prospective relative-risk scope and selected core variables including curvature, friction, visibility, and speed.
Stage 2
OSM Integration, Segmentation, and Map Selection
Integrated OpenStreetMap-derived data, road segmentation, coordinate extraction, and an interactive map-selection interface.
Stage 3
Heading Change, Radius, Curvature, and Bend Metrics
Converted coordinate sequences into heading change, radius of curvature, curvature scoring, and bend-frequency context.
Stage 4
Centripetal Acceleration, Friction, Slip, and Safe Speed
Added kinematic and dynamic relationships linking road geometry with lateral demand, grip assumptions, and safe-speed estimates.
Stage 5
Stopping Distance, Reaction Time, and Vehicle Profiles
Introduced reaction distance, velocity-squared braking behaviour, vehicle parameters, and profile-dependent interpretation.
Stage 6
Road Context, Surface, Lighting, and Infrastructure Factors
Added road classification, traffic proxy, lane context, surface condition, lighting, pedestrian/cycle infrastructure, and barrier effects.
Stage 7
Environmental and Behavioural Scenario Modelling
Integrated rain, fog, ice, snow, flooding, overspeed, fatigue, distraction, BAC, and combined multiplier assumptions.
Stage 8
Risk Aggregation, Normalisation, Graphs, Routes, and Exports
Integrated geometry, physics, context, environment, and behaviour into a normalised comparative output with graphs, route analysis, and exports.
Stage 9
Public Research-Site Framing
Refactored the surrounding website to explain the model, limitations, references, results structure, and live-app workflow.
Technical Architecture
Public Web Technologies and Open Data Services
Map Interface
Leaflet and MapTiler
The live app uses a browser map interface for road selection, overlays, routes, isochrones, and visual context.
Road Data
OpenStreetMap and Overpass
Road ways, tags, and geometry are retrieved from public map data where available.
Routing
OpenRouteService
Route and isochrone functions use external routing services while the app handles map display and risk sampling.
Maths and Reporting
MathJax, Graphs, and Exports
The app exposes formula explanations, distribution views, and export files so the model can be inspected.
Methodology
Traceable Model Construction
The live app connects map selection, geometry extraction, physics calculations, scenario multipliers, statistical context, and exports. That makes the system easier to inspect than a single untraceable number shown without provenance.
The public website now acts as the research companion to the app: it explains the method, formulas, assumptions, limitations, and intended use before users launch the interactive model.
Method Overview
- Select
- Choose a road segment from the map.
- Derive
- Calculate geometry and physics values.
- Assume
- Apply transparent scenario and fallback assumptions.
- Compare
- Use graphs, route sampling, and percentile context.
- Export
- Preserve results for review and communication.
- Review
- Add only reviewed case-evidence JSON to the public Results dataset.
Testing and Validation Status
Implementation Status and Validation Boundary
Checked So Far
- Public pages link to the stable live app.
- Core app outputs are designed to expose maths, graphs, and exports.
- Scenario and limitation wording is visible on the public site.
- The model is framed as comparative rather than official prediction.
Validation boundary
The project does not yet claim official calibration against national crash datasets, formal engineering audits, or government road-safety standards. Future work should compare model outputs against trusted external evidence.
Data Sources
Public Data Sources and Limitations
Source Categories
OpenStreetMap
Overpass API
OpenRouteService
MapTiler
Derived geometry
Scenario assumptions
Road Risk benefits from public data because it is transparent and widely available. It also inherits public-data limitations such as missing tags, inconsistent geometry density, and incomplete infrastructure attributes.
Data limitation
Missing public tags must not be treated as proof that a feature is absent. The app uses fallback assumptions and confidence notes where possible, but stronger conclusions require better data or field verification.
Data Pipeline Detail
Retrieved, Derived, Assumed, and Exported Data
01
Retrieved
OSM ways, nodes, tags, road class, names/refs, speed/surface hints, and public map context.
02
Derived
Selected segment, bearing, heading change, curvature, radius, bounding context, safe speed, and stopping distance.
03
Assumed
Vehicle profile, weather, visibility, behaviour, friction, traffic proxy, and missing-data fallbacks.
04
Exported
CSV, GeoJSON, JSON run summaries, distribution data, and visual/map-style outputs for review.
For Judges
Fast Review Route
In 30 seconds: Road Risk tests whether public road geometry and physics-informed assumptions can produce transparent, prospective relative-risk comparisons. In two minutes: open the live app, select a road, inspect Annualised Comparative Model Output, safe speed and stopping distance, change one scenario setting, then compare the maths or graph view.
Launch
Open the live app from the header or CTA.
Select
Click a road and wait for the selected-road output.
Interpret
Read the output as comparative, not official.
Stress-Test
Change rain, fog, fatigue, vehicle type, or overspeed.
Judge Questions
What Judges Should Ask Me
Meaning
What Does the Output Mean?
It is a comparative model output under defined assumptions, not a measured crash rate.
Data Choice
Why Not Use Crash Data Yet?
The project first tests whether geometry and physics can produce a transparent prospective signal before calibration.
Sensitivity
How Do Assumptions Affect Results?
Change one factor at a time, such as rain, fog, fatigue, or overspeed, and compare the same selected road.
Mechanism
How Does Curvature Influence the Model?
Tighter curvature and smaller radius increase lateral demand and can reduce the friction-limited safe-speed estimate.
Next Validation
How Could This Be Validated Next?
Compare exported cases with official collision records, engineering audits, field inspection, and expert review.
Evidence Trail
Where the Evidence Lives
Physical Display
Booklet and poster materials are presented as SciFest display resources, not public downloads.
Live App
Selected-road calculations, scenario controls, graphs, maths panels, routes, and exports are inspectable.
Results Cases
Local cases appear on the current browser; reviewed cases can be committed to the public static dataset.
Exports
CSV, JSON, GeoJSON, graph data, and case-evidence files provide reviewable records.
References
Sources support the method and context; they do not endorse the app or validate individual outputs.
Responsible Use
What Not to Conclude
It may indicate a segment worth closer review under the active assumptions.
The model cannot see every real-world condition, behaviour, or infrastructure defect.
Fallback assumptions are used where public tags are incomplete.
Operational use would require calibration, field verification, and professional assessment.
Research Framing
Research Framing for the Development Process
Explain the Investigation
The project should be presented as a research process: question, method, model, evidence, limitations, and future work.
Document Iteration Transparently
Development logs and tracked versions help show how the model, app, and public presentation evolved over time.
Support Technical Audit
Keeping the source inspectable matters because the project depends on transparent assumptions and reproducible calculations.
External Feedback
Research Context and Review Direction
External discussion and reference gathering are treated as research context, not endorsement or validation. The next stage is to review assumptions with road-safety, transport, and vehicle-dynamics expertise.
Expert Review Pathway
The project framing identifies possible review routes with academic transport researchers, road-safety organisations, and public-sector infrastructure specialists.
Validation Requirement
Stronger claims would require comparison against observed collision datasets, engineering assessments, road-condition surveys, and documented infrastructure context.
Future Work
Future Development and Research Direction
Validation
Compare with External Evidence
Use crash datasets, audit examples, known blackspots, or expert review to test model behaviour.
Data Quality
Improve Data-Confidence Flags
Make missing tags, fallback assumptions, geometry sparsity, and route simplification even easier to inspect.
Architecture
Modularise Safely
Separate public site, model helpers, map logic, API helpers, graphing, and exports without breaking the stable app.
Deployment
Harden API Use
Longer-term, route protected API calls through a server-side proxy with caching, rate limits, and clearer failure states.
External Feedback
Review Assumptions with Domain Expertise
Road-safety, transport, education, and GIS feedback would help separate strong model components from areas needing better evidence.
Current Build
The Live Application Remains the Working Model
The public site explains the project. The existing live app remains unchanged and available for road selection, route analysis, graphs, maths, and exports.