Abstract

(à rédiger)

1. Introduction

2. Architecture RS3→Telemachus

3. Curvature-aware modules

4. Datasets & Reproducibility

5. Case Studies

6. Roadmap v1.0

References

id: P009 title: “Curvature-Aware Simulation Pipelines for Mobility and Safety Research” status: drafting theme: curvature venue: “White Paper / arXiv (2025)” tags: [curvature, simulation, RS3, telemachus, road-geometry, ADAS]

Abstract

This white paper introduces a curvature-aware simulation pipeline for vehicle mobility and safety research, combining high-fidelity inertial simulation (RS3) with open mobility data standards (Telemachus).
The objective is to establish curvature as a first-class citizen in trajectory-level simulation and to connect geometric realism with risk and comfort metrics.
Our pipeline bridges the gap between physics-based modeling, open data validation, and reproducibility, paving the way for industrial and academic collaboration in the analysis of road geometry and driver assistance systems.

1. Introduction

Curvature defines the geometry of motion. In road mobility, it directly governs lateral acceleration, steering control, and perception risk.
However, existing simulation pipelines often simplify curvature as static map data rather than a dynamic variable influenced by speed, inertia, and terrain grade.
The RS3–Telemachus framework aims to restore curvature as a core analytical feature, linking synthetic and real trajectories in a reproducible open-source workflow.

2. Architecture: RS3 → Telemachus

The RS3 simulator generates 10 Hz inertial trajectories (GNSS, IMU, wheel speed, altitude) with realistic noise and delay models.
Each dataset follows the Telemachus 1.0 schema, ensuring interoperability across research tools.
Data are exported as telemachus-py validated records, then ingested into analysis modules for curvature, slope, and dynamic stability computation.
A modular architecture allows for plug-in integration of map-matching, risk assessment, and event-detection algorithms.

RS3 Simulation Engine → Telemachus Exporter → Curvature & Grade Estimator → Risk Scoring → Dataset Publication

3. Curvature-Aware Modules

The curvature-aware pipeline includes three principal modules:

3.1. Curvature Estimation

A spline-based estimator computes continuous κ(s) and its derivative κ’(s), providing high-resolution profiles of turning geometry.
Filtering ensures smooth transitions across segments while preserving local curvature extremes critical for driver dynamics.

3.2. Dynamic Consistency

Couples curvature with grade and yaw rate to evaluate stability envelopes: [ a_y = v^2 κ, \quad a_x = g \sin(θ) ] These metrics quantify comfort, maneuverability, and lateral load transfer under varying conditions.

3.3. Risk and Comfort Index

An interpretable risk index is computed: [ R = α|κ| + β|\tan(θ)| + γ(v/v_{max})^2 ] Empirically validated against curvature-induced accident datasets and simulated driver reactions in RS3.

4. Datasets & Reproducibility

All curvature-aware datasets adhere to the Telemachus dataset contract, including metadata (source, uncertainty, context).
Reproducibility is achieved through versioned simulation campaigns, open code (GitHub: telemachus3 / roadsimulator3), and documented validation scripts.
Integration with public archives (Zenodo, Kaggle) ensures long-term accessibility.

5. Case Studies

• RS3 Curvature Benchmark: comparison between simulated curvature and OSM/IGN road centerlines.
• Risk Simulation: quantifying curvature-related discomfort under variable speeds and grades.
• ADAS Evaluation: using curvature dynamics to test lane-keeping assist (LKA) robustness.

Each case study demonstrates the capacity of the RS3–Telemachus pipeline to translate geometric insight into actionable safety indicators.

6. Roadmap v1.0

• Q1 2025: Release RS3 curvature dataset v1.0 and validation scripts.
• Q2 2025: Publish Telemachus “Road Genome” RFC defining curvature and grade metadata.
• Q3 2025: Submit results to IEEE IV or Sensors.
• Q4 2025: Integrate curvature-aware risk modeling into RS3 core.

References

• Drosescu A. et al., Road Curvature Measurement Using IMU/GPS, ACME 2016.
• IFAC 2023, Efficient Real‑Time Road Curvature Estimation – Visual–Inertial Approach.
• US Patent 7522091, Road Curvature Estimation System.
• Road Geometry Risk Estimation Model, 2020.
• Telemachus Specification 1.0, Open Mobility Data Standard (2025).
• RoadSimulator3, Open Simulation Framework for Inertial Trajectories.