How Do We Fail? Stress Testing Perception in Autonomous Vehicles

Delecki, Harrison; Itkina, Masha; Lange, Bernard; Senanayake, Ransalu; Kochenderfer, Mykel J.

doi:10.1109/iros47612.2022.9981724

Cited by 10 publications

(3 citation statements)

References 33 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Piroli et al [69] detected the presence of rain and snow using an energy-based anomaly detection framework. Li et al [70] evaluated and modeled LiDAR visibility under different artificial fog conditions, while Delecki et al [71] increased pressure on the recognition model by gradually adding computer-synthesized rain, snow, and fog to analyze the causes of recognition failures.…”

Section: Three-dimensional Environmental Perception In Adverse Weathermentioning

confidence: 99%

Multi-Modal Contrastive Learning for LiDAR Point Cloud Rail-Obstacle Detection in Complex Weather

Wen,

Peng,

Lin

et al. 2024

Electronics

View full text Add to dashboard Cite

Obstacle intrusion is a serious threat to the safety of railway traffic. LiDAR point cloud 3D semantic segmentation (3DSS) provides a new method for unmanned rail-obstacle detection. However, the inevitable degradation of model performance occurs in complex weather and hinders its practical application. In this paper, a multi-modal contrastive learning (CL) strategy, named DHT-CL, is proposed to improve point cloud 3DSS in complex weather for rail-obstacle detection. DHT-CL is a camera and LiDAR sensor fusion strategy specifically designed for complex weather and obstacle detection tasks, without the need for image input during the inference stage. We first demonstrate how the sensor fusion method is more robust under rainy and snowy conditions, and then we design a Dual-Helix Transformer (DHT) to extract deeper cross-modal information through a neighborhood attention mechanism. Then, an obstacle anomaly-aware cross-modal discrimination loss is constructed for collaborative optimization that adapts to the anomaly identification task. Experimental results on a complex weather railway dataset show that with an mIoU of 87.38%, the proposed DHT-CL strategy achieves better performance compared to other high-performance models from the autonomous driving dataset, SemanticKITTI. The qualitative results show that DHT-CL achieves higher accuracy in clear weather and reduces false alarms in rainy and snowy weather.

show abstract

Section: Three-dimensional Environmental Perception In Adverse Weathermentioning

confidence: 99%

Multi-Modal Contrastive Learning for LiDAR Point Cloud Rail-Obstacle Detection in Complex Weather

Wen,

Peng,

Lin

et al. 2024

Electronics

View full text Add to dashboard Cite

show abstract

“…In [28], researchers utilize a GAN model to produce adversarial objects that can be used to attack LiDAR-based driving systems. In [29], the authors propose a methodology for performing stress tests on LiDAR-based perception. The researchers use a real-world driving dataset and various weather conditions to evaluate the performance of autonomous driving systems.…”

Section: A Adversarial Noise Attacksmentioning

confidence: 99%

Adversarial Deep Reinforcement Learning for Improving the Robustness of Multi-agent Autonomous Driving Policies

Sharif

Marijan

2022

2022 29th Asia-Pacific Software Engineering Conference (APSEC)

View full text Add to dashboard Cite

Autonomous vehicles are advanced driving systems that are well known for being vulnerable to various adversarial attacks, compromising the vehicle's safety, and posing danger to other road users. Rather than actively training complex adversaries by interacting with the environment, there is a need to first intelligently find and reduce the search space to only those states where autonomous vehicles are found less confident. In this paper, we propose a blackbox testing framework ReMAV using offline trajectories first to analyze the existing behavior of autonomous vehicles and determine appropriate thresholds for finding the probability of failure events. To this end, we introduce a three-step methodology which i) uses offline state action pairs of any autonomous vehicle under test, ii) builds an abstract behavior representation using our designed reward modeling technique in order to analyze states with uncertain driving decisions, and iii) uses a disturbance model for minimal perturbation attacks where the driving decisions are less confident. Our reward modeling technique helps in creating a behavior representation that allows us to highlight regions of likely uncertain behavior even when the baseline autonomous vehicle is performing well. This approach allows for more efficient testing without the need for computational and inefficient active adversarial learning techniques. We perform our experiments in a high-fidelity urban driving environment using three different driving scenarios containing single and multi-agent interactions. Our experiment shows 35%, 23%, 48%, and 50% increase in occurrences of vehicle collision, road objects collision, pedestrian collision, and offroad steering events respectively by the autonomous vehicle under test, demonstrating a significant increase in failure events. We also perform a comparative analysis with prior testing frameworks and show that they underperform in terms of training-testing efficiency, finding total infractions, and simulation steps to identify the first failure compared to our approach. The results show that the proposed framework can be used to understand existing weaknesses of the autonomous vehicles under test in order to only attack those regions, starting with the simplistic perturbation models. We demonstrate that our approach is capable of finding failure events in terms of increased collision and offroad steering errors compared to the baseline driving behavior.

show abstract

“…The performance evaluation of automotive data fusion often requires summarizing aspects of perceptual performance into a small number of scalar values for comparison [33]. Computing lower-level metrics requires associating the estimated tracks of the System-Under-Test to their corresponding reference tracks, which is realized in the following way: the pairwise distances between all estimated and all reference objects or tracks are computed using an object distance function [34,35]. In this paper, we use the Euclidean distance τ of the observation point of the MMW radar and the center point of the camera sensor to associate them and then use the ratio of τ to the vehicle length W as the threshold for fault classification.…”

mentioning

confidence: 99%

Fault Diagnosis of the Autonomous Driving Perception System Based on Information Fusion

Hou

Li³

2023

Sensors

View full text Add to dashboard Cite

The reliability of autonomous driving sensing systems impacts the overall safety of the driving system. However, perception system fault diagnosis is currently a weak area of research, with limited attention and solutions. In this paper, we present an information-fusion-based fault-diagnosis method for autonomous driving perception systems. To begin, we built an autonomous driving simulation scenario using PreScan software, which collects information from a single millimeter wave (MMW) radar and a single camera sensor. The photos are then identified and labeled via the convolutional neural network (CNN). Then, we fused the sensory inputs from a single MMW radar sensor and a single camera sensor in space and time and mapped the MMW radar points onto the camera image to obtain the region of interest (ROI). Lastly, we developed a method to use information from a single MMW radar to aid in diagnosing defects in a single camera sensor. As the simulation results show, for missing row/column pixel failure, the deviation typically falls between 34.11% and 99.84%, with a response time of 0.02 s to 1.6 s; for pixel shift faults, the deviation range is between 0.32% and 9.92%, with a response time of 0 s to 0.16 s; for target color loss, faults have a deviation range of 0.26% to 2.88% and a response time of 0 s to 0.05 s. These results prove the technology is effective in detecting sensor faults and issuing real-time fault alerts, providing a basis for designing and developing simpler and more user-friendly autonomous driving systems. Furthermore, this method illustrates the principles and methods of information fusion between camera and MMW radar sensors, establishing the foundation for creating more complicated autonomous driving systems.

show abstract

How Do We Fail? Stress Testing Perception in Autonomous Vehicles

Cited by 10 publications

References 33 publications

Multi-Modal Contrastive Learning for LiDAR Point Cloud Rail-Obstacle Detection in Complex Weather

Multi-Modal Contrastive Learning for LiDAR Point Cloud Rail-Obstacle Detection in Complex Weather

Adversarial Deep Reinforcement Learning for Improving the Robustness of Multi-agent Autonomous Driving Policies

Fault Diagnosis of the Autonomous Driving Perception System Based on Information Fusion

Contact Info

Product

Resources

About