NeuroNCAP: Photorealistic Closed-loop Safety Testing for Autonomous Driving

TL;DR

NeuroNCAP introduces a NeRF-based simulator for realistic closed-loop testing of autonomous driving systems, revealing critical flaws in current models under safety-critical scenarios and providing a new tool for research validation.

Contribution

The paper presents a novel NeRF-based simulator for autonomous driving, enabling realistic, reconfigurable, safety-critical scenario testing and benchmarking of AD models.

Findings

State-of-the-art end-to-end planners perform well in nominal scenarios

Critical flaws are revealed in safety-critical scenarios in closed-loop testing

The simulator and scenarios are publicly released for community use

Abstract

We present a versatile NeRF-based simulator for testing autonomous driving (AD) software systems, designed with a focus on sensor-realistic closed-loop evaluation and the creation of safety-critical scenarios. The simulator learns from sequences of real-world driving sensor data and enables reconfigurations and renderings of new, unseen scenarios. In this work, we use our simulator to test the responses of AD models to safety-critical scenarios inspired by the European New Car Assessment Programme (Euro NCAP). Our evaluation reveals that, while state-of-the-art end-to-end planners excel in nominal driving scenarios in an open-loop setting, they exhibit critical flaws when navigating our safety-critical scenarios in a closed-loop setting. This highlights the need for advancements in the safety and real-world usability of end-to-end planners. By publicly releasing our simulator and scenarios as an easy-to-run evaluation suite, we invite the research community to explore, refine, and validate their AD models in controlled, yet highly configurable and challenging sensor-realistic environments. Code and instructions can be found at https://github.com/atonderski/neuro-ncap