Abstract
Ensuring the safety of Vulnerable Road Users (VRUs) is a primary goal for Advanced Driver Assistance Systems (ADAS). This paper presents a lightweight, configurable simulation framework designed to digitize Euro NCAP safety protocols and evaluate interactions between vehicles and diverse VRUs, including pedestrians, e-scooters, and motorcyclists. By integrating Vehicle-to-Everything (V2X) communication with a rule-based finite-state machine motion planner, the framework enables a holistic assessment of safety, traffic efficiency, and passenger comfort. Experimental results demonstrate that V2X-assisted systems can provide a critical "one-second advantage," preventing collisions in high-risk scenarios where traditional onboard sensors fail, while also improving passenger comfort through smoother deceleration.
Introduction
As urban mobility landscapes shift, the category of Vulnerable Road Users (VRUs) has expanded beyond pedestrians and cyclists to include electric scooter (e-scooter) riders and motorcyclists. Testing these interactions in the real world is prohibitively expensive, dangerous, and difficult to replicate. While simulation is the obvious solution, existing platforms often lack diversity in VRU types, suffer from rigid architectures that make it difficult to swap algorithms, and rarely provide a holistic view that balances safety, efficiency, and comfort. This paper addresses these gaps by proposing a modular, three-stage simulation framework.
Framework Architecture
The proposed framework addresses these gaps through a three-stage modular design: Scenario Design, Environment Modeling, and Simulation Analysis.
- Scenario Design: Standardized testing scenarios are decomposed into functional components and specific Operational Design Domains (ODDs).
- Environment Modeling: The framework models VRUs as kinematic point masses and the ego-vehicle using a kinematic bicycle model. This approach maintains computational efficiency while allowing for diverse agent types.
- Simulation Analysis: The system validates scenarios against specific ODDs and evaluates performance using a multi-dimensional metric.
Motion Planning and V2X Integration
The ego-vehicle's "brain" is a rule-based Finite-State Machine (FSM) that manages transitions between three states: Cruise (maintaining desired speed), Pre-decelerate (gentle preventive braking), and Emergency Brake.
A key feature of the framework is the integration of "idealized" V2X communication. The vehicle receives real-time state information (position and velocity) from VRUs or roadside infrastructure. This data is fused with onboard sensor data—simulated as a 120° field-of-view sector—to enhance situational awareness beyond the line of sight. Safety logic is governed by a "two-second rule" to define safe following distances ($D_{safe} = 2 \cdot V_{veh}$).
Evaluation Scenarios
The framework digitizes Euro NCAP VRU testing protocols to create a synthetic benchmark. Three primary scenarios were modeled:
- Car-to-Pedestrian Nearside Adult: A pedestrian crossing a straight road.
- Car-to-Bicyclist Longitudinal Adult: Adapted here for an e-scooter leading the vehicle in-lane.
- Car-to-Motorcyclist Front Turn across Path: A complex intersection conflict during a vehicle turn.
Results and Discussion
The framework was validated using a "Sustainability Score" that aggregates Safety, Efficiency, and Comfort. The table below summarizes the performance across the three digitized Euro NCAP scenarios.
| Scenario | Controller | Safety Score (0-1) | Comfort Score (0-1) |
|---|---|---|---|
| Pedestrian Crossing | Baseline (Sensors Only) | 0.00 (Collision) | 0.00 |
| V2X-Assisted | 1.00 (Safe) | 0.59 | |
| Motorcyclist Crossing | Baseline (Sensors Only) | 0.00 (Collision) | 0.00 |
| V2X-Assisted | 0.81 (Safe) | 0.89 | |
| E-scooter Following | Baseline (Sensors Only) | 1.00 | 0.78 |
| V2X-Assisted | 1.00 | 0.76 |
As shown in the pedestrian crossing simulation above, the baseline controller fails to detect the VRU in time, resulting in a collision at $t=5.00s$. In contrast, the V2X-assisted system initiates braking earlier, allowing for a safe stop.
The "One-Second" Advantage
In scenarios with orthogonal conflicts, V2X communication shifts the braking transition forward by roughly one second. This margin is often the difference between a fatal collision and a safe stop. Furthermore, V2X allows for earlier, gentler deceleration in following scenarios (e.g., e-scooters), significantly improving passenger comfort by reducing jerk, even if it slightly reduces overall traffic efficiency due to more conservative behavior.
Conclusion
This configurable framework provides a robust tool for assessing VRU safety in a virtual environment. While the current results assume idealized V2X communication, future work will focus on introducing realistic communication effects like latency and packet loss. Additionally, the authors plan to upgrade the motion planner to advanced predictive models, such as Model Predictive Control (MPC), to handle more complex multi-agent interactions.
Summary updated