Abstract:Autonomous Vehicles (AVs) have the potential to revolutionize transportation systems by enhancing traffic safety. Safety testing is undoubtedly a critical step for enabling large-scale deployment of AVs. High-risk scenarios are particularly important as they pose significant challenges and provide valuable insights into the driving capabilities of AVs. This study presents a novel approach to assess the safety of AVs using in-depth crash data, with a particular focus on real-world crash scenarios. First, based on the high-definition video recording of the whole process prior to the crash occurrences, 453 real-world crashes involving 596 passenger cars from China In-depth Mobility Safety Study-Traffic Accident (CIMSS-TA) database were reconstructed. Pertinent static and dynamic elements needed for the construction of the testing scenarios were extracted. Subsequently, 596 testing scenarios were created via each passenger car's perspective within the simulation platform. Following this, each of the crash-involved passenger cars was replaced with Baidu Apollo, a famous automated driving system (ADS), for counterfactual simulation. Lastly, the safety performance of the AV was assessed using the simulation results. A logit model was utilized to identify the fifteen crucial scenario elements that have significant impacts on the test results. The findings demonstrated that the AV could avoid 363 real-world crashes, accounting for approximately 60.91% of the total, and effectively mitigated injuries in the remaining 233 unavoidable scenarios compared to a human driver. Moreover, the AV maintain a smoother speed in most of the scenarios. The common feature of these unavoidable scenarios is that the AV is in a passive state, and the crashes are not caused by the AV violating traffic rules, but rather caused by abnormal behavior exhibited by the human drivers. Additionally, seven specific scenarios have been identified wherein AVs are unable to avoid a crash. These findings demonstrate that, compared to human drivers, AVs can avoid crashes that are difficult for humans to avoid, thereby enhancing traffic safety.

DeepCrashTest: Turning Dashcam Videos into Virtual Crash Tests for Automated Driving Systems

From Dashcam Videos to Driving Simulations: Stress Testing Automated Vehicles against Rare Events

A system of vision sensor based deep neural networks for complex driving scene analysis in support of crash risk assessment and prevention

DeepTest: Automated Testing of Deep-Neural-Network-driven Autonomous Cars

CRASH: Crash Recognition and Anticipation System Harnessing with Context-Aware and Temporal Focus Attentions

CycleCrash: A Dataset of Bicycle Collision Videos for Collision Prediction and Analysis

Review of Accident Detection Methods Using Dashcam Videos for Autonomous Driving Vehicles

Edge Computing for Real-Time Near-Crash Detection for Smart Transportation Applications

CRASH: Challenging Reinforcement-Learning Based Adversarial Scenarios For Safety Hardening

A Digital Twin-based Testing and Data Collection System for Autonomous Driving in Extreme Traffic Scenarios

How would autonomous vehicles behave in real-world crash scenarios?

Computer Vision based Accident Detection for Autonomous Vehicles

A Gaze Data-based Comparative Study to Build a Trustworthy Human-AI Collaboration in Crash Anticipation

Automated Vehicle Crash Sequences: Patterns and Potential Uses in Safety Testing

Unsupervised Traffic Accident Detection in First-Person Videos

Vehicle crash simulations for safety: Introduction of connected and automated vehicles on the roadways

Automatic Map Update Using Dashcam Videos

Enhanced Transfer Learning for Autonomous Driving with Systematic Accident Simulation

In-Vehicle Edge System for Real-Time Dashcam Video Analysis

Simulation of Urban Crash Occurrence Based on Real-World Crash Data