Evaluation of Pedestrian Safety in a High-Fidelity Simulation Environment Framework

TL;DR

This work addresses pedestrian safety evaluation in autonomous driving by integrating pre-crash conflicts and post-crash injury severity into a unified estimator within a Carla-based digital twin framework. It reconstructs environment, traffic, and pedestrian characteristics and enables in-loop cooperative perception between roadside and vehicle sensors, using a logistic injury model dependent on $V$ and $A$ to quantify safety outcomes. Experiments at a crowded Beijing intersection show that V2I cooperative perception substantially reduces conflicts, collisions, and injuries compared with single-vehicle perception, validating the framework's ability to compare pedestrian safety across algorithms. The proposed platform serves as a practical, configurable tool for standardized pedestrian safety assessment and design of more pedestrian-friendly autonomous driving strategies.

Abstract

Pedestrians' safety is a crucial factor in assessing autonomous driving scenarios. However, pedestrian safety evaluation is rarely considered by existing autonomous driving simulation platforms. This paper proposes a pedestrian safety evaluation method for autonomous driving, in which not only the collision events but also the conflict events together with the characteristics of pedestrians are fully considered. Moreover, to apply the pedestrian safety evaluation system, we construct a high-fidelity simulation framework embedded with pedestrian safety-critical characteristics. We demonstrate our simulation framework and pedestrian safety evaluation with a comparative experiment with two kinds of autonomous driving perception algorithms -- single-vehicle perception and vehicle-to-infrastructure (V2I) cooperative perception. The results show that our framework can evaluate different autonomous driving algorithms with detailed and quantitative pedestrian safety indexes. To this end, the proposed simulation method and framework can be used to access different autonomous driving algorithms and evaluate pedestrians' safety performance in future autonomous driving simulations, which can inspire more pedestrian-friendly autonomous driving algorithms.

Figures (6)

• Figure 1: The real-world and simulated scenario of a crowded urban intersection using the proposed high fidelity simulation framework.
• Figure 2: The proposed high fidelity simulation framework based on Carla.
• Figure 3: Schematic diagram of Scenarios. a.Crossing. b.Jaywalking. c.Right turning. The blue circle represents the roadside camera.
• Figure 4: The injury severity curve of single vehicle and V2I condition in the jaywalking scenario
• Figure 5: Human-vehicle distance distribution when the sensor module detected the pedestrian for the first time.