Evidential deep learning for arbitrary LIDAR object classification in the context of autonomous driving

Capellier, Edouard; Davoine, Franck; Cherfaoui, Véronique; You, Li

doi:10.1109/ivs.2019.8813846

Cited by 29 publications

(12 citation statements)

References 24 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…There are also content layer corner case detection methods on LiDAR data, such as Wong et al [26], who introduce an open-set instance segmentation network on point clouds that identifies unknown points in an embedding space and groups them into unknown instances. Capellier et al [27] propose a method to detect both known and unknown objects in LiDAR data. Bolte et al [9] identify temporal layer corner cases for the camera sensor if the residual error between the real and a predicted image, weighted by the criticality of the location in the image, exceeds a threshold.…”

Section: Corner Case Detection Techniquesmentioning

confidence: 99%

An Application-Driven Conceptualization of Corner Cases for Perception in Highly Automated Driving

Heidecker,

Breitenstein,

Rösch

et al. 2021

Preprint

View full text Add to dashboard Cite

Systems and functions that rely on machine learning (ML) are the basis of highly automated driving. An essential task of such ML models is to reliably detect and interpret unusual, new, and potentially dangerous situations. The detection of those situations, which we refer to as corner cases, is highly relevant for successfully developing, applying, and validating automotive perception functions in future vehicles where multiple sensor modalities will be used. A complication for the development of corner case detectors is the lack of consistent definitions, terms, and corner case descriptions, especially when taking into account various automotive sensors. In this work, we provide an application-driven view of corner cases in highly automated driving. To achieve this goal, we first consider existing definitions from the general outlier, novelty, anomaly, and out-of-distribution detection to show relations and differences to corner cases. Moreover, we extend an existing camera-focused systematization of corner cases by adding RADAR (radio detection and ranging) and LiDAR (light detection and ranging) sensors. For this, we describe an exemplary toolchain for data acquisition and processing, highlighting the interfaces of the corner case detection. We also define a novel level of corner cases, the method layer corner cases, which appear due to uncertainty inherent in the methodology or the data distribution.

show abstract

Section: Corner Case Detection Techniquesmentioning

confidence: 99%

An Application-Driven Conceptualization of Corner Cases for Perception in Highly Automated Driving

Heidecker,

Breitenstein,

Rösch

et al. 2021

Preprint

View full text Add to dashboard Cite

show abstract

“…We previously observed that using Instance‐Normalization (Ulyanov et al, 2016) in the final layer of the classifier would be an easy way to solve those problems (Capellier et al, 2019). Let

υ (x) = (υ_{1} (x), …, υ_{d} (x))

be the mapping modeled by all the consecutive layers of the classifier but the last one; let

\overset{υ_{j}}{true}

be the mean value of the

υ_{j}

function on the training set, and

σ {MathClass-open(υ_{j} MathClass-close)}^{2}

its corresponding variance.…”

Section: Generation Of Evidential Mass Functions From a Binary Glr Classifier Trained To Detect The Road In A Lidar Scanmentioning

confidence: 99%

Fusion of neural networks, for LIDAR‐based evidential road mapping

Capellier

Davoine

Cherfaoui

et al. 2021

Journal of Field Robotics

Self Cite

View full text Add to dashboard Cite

LIDAR sensors are usually used to provide autonomous vehicles with three‐dimensional representations of their environment. In ideal conditions, geometrical models could detect the road in LIDAR scans, at the cost of a manual tuning of numerical constraints, and a lack of flexibility. We instead propose an evidential pipeline, to accumulate road detection results obtained from neural networks. First, we introduce RoadSeg, a new convolutional architecture that is optimized for road detection in LIDAR scans. RoadSeg is used to classify individual LIDAR points as either belonging to the road, or not. Yet, such point‐level classification results need to be converted into a dense representation, that can be used by an autonomous vehicle. We thus second present an evidential road mapping algorithm, that fuses consecutive road detection results. We benefitted from a reinterpretation of logistic classifiers, which can be seen as generating a collection of simple evidential mass functions. An evidential grid map that depicts the road can then be obtained, by projecting the classification results from RoadSeg into grid cells, and by handling moving objects via conflict analysis. The system was trained and evaluated on real‐life data. A python implementation maintains a 10 Hz framerate. Since road labels were needed for training, a soft labeling procedure, relying lane‐level HD maps, was used to generate coarse training and validation sets. An additional test set was manually labeled for evaluation purposes. So as to reach satisfactory results, the system fuses road detection results obtained from three variants of RoadSeg, processing different LIDAR features.

show abstract

“…Lidars provide a good physical description of the target and, due to that, lidars have been used for target detection, tracking, and motion prediction., Filtering of the ground and clustering of the target [ 102 , 103 , 104 , 105 ] are two methods widely used by lidar for object detection, which provide the spatial information of the target. To classify and recognize objects (like pedestrians, trees, or vehicles), lidars make use of techniques such as machine learning based on object recognition [ 106 , 107 , 108 , 109 ], and additional methods such as global and local extraction of features to help in providing the structure of the target. Lidar uses the Bayesian filtering framework and data association methods for target tracking and motion prediction to provide information, such as velocity, trajectory, and object positioning [ 110 , 111 , 112 ].…”

Section: Sensorsmentioning

confidence: 99%

The Perception System of Intelligent Ground Vehicles in All Weather Conditions: A Systematic Literature Review

Mohammed

Amamou

Ayevide

et al. 2020

Sensors

View full text Add to dashboard Cite

Perception is a vital part of driving. Every year, the loss in visibility due to snow, fog, and rain causes serious accidents worldwide. Therefore, it is important to be aware of the impact of weather conditions on perception performance while driving on highways and urban traffic in all weather conditions. The goal of this paper is to provide a survey of sensing technologies used to detect the surrounding environment and obstacles during driving maneuvers in different weather conditions. Firstly, some important historical milestones are presented. Secondly, the state-of-the-art automated driving applications (adaptive cruise control, pedestrian collision avoidance, etc.) are introduced with a focus on all-weather activity. Thirdly, the most involved sensor technologies (radar, lidar, ultrasonic, camera, and far-infrared) employed by automated driving applications are studied. Furthermore, the difference between the current and expected states of performance is determined by the use of spider charts. As a result, a fusion perspective is proposed that can fill gaps and increase the robustness of the perception system.

show abstract

Evidential deep learning for arbitrary LIDAR object classification in the context of autonomous driving

Cited by 29 publications

References 24 publications

An Application-Driven Conceptualization of Corner Cases for Perception in Highly Automated Driving

An Application-Driven Conceptualization of Corner Cases for Perception in Highly Automated Driving

Fusion of neural networks, for LIDAR‐based evidential road mapping

The Perception System of Intelligent Ground Vehicles in All Weather Conditions: A Systematic Literature Review

Contact Info

Product

Resources

About