CaT: CAVS Traversability Dataset for Off-Road Autonomous Driving

Sharma, Suvash; Dabbiru, Lalitha; Hannis, Tyler; Mason, George L.; Carruth, Daniel W.; Doude, Matthew; Goodin, Chris; Hudson, Christopher R.; Ozier, Sam; Ball, John; Tang, Bo

doi:10.1109/access.2022.3154419

Cited by 26 publications

(13 citation statements)

References 21 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…On the other hand, off-road environments contain challenging features such as changing terrain types and vegetation with complex geometry. Only a few datasets exists for off-road environments, which are RELLIS-3D, 10 ORFD, 5 and CaT; 30 and out of these, only RELLIS-3D offers labeled LiDAR data. Due to the challenge in annotating LiDAR datasets, some studies rely on simulations to generate labeled data.…”

Section: Lidar Datasetsmentioning

confidence: 99%

Analysis of LiDAR configurations on off-road semantic segmentation performance

Chen

Dabbiru

et al. 2023

Autonomous Systems: Sensors, Processing and Security for Ground, Air, Sea, and Space Vehicles and Infrastructure 2023

View full text Add to dashboard Cite

LiDAR-based 3D semantic segmentation is one of the most widely used perception methods to support scene understanding of self-driving vehicles. Most publicly available LiDAR datasets for driving scene segmentation, such as SemanticKITTI, nuScenes, and SemanticPOSS, provide only a single type of LiDAR configuration. Therefore, testing a trained model with a different channel configuration than the training dataset is sometimes inevitable in real-world applications. Despite the significance of this LiDAR channel mismatch problem in the machine learning pipeline, little research has focused on investigating the impact of the LiDAR configuration shift on a model's test performance. This paper aims to provide referenceable baseline experiments for the LiDAR configuration shifts. We explore the effect of using different LiDAR channels when training and testing a 3D LiDAR point cloud semantic segmentation model, utilizing Cylinder3D for the experiments. A Cylinder3D model is trained and tested on simulated 3D LiDAR point cloud datasets created using the Mississippi State University Autonomous Vehicle Simulator (MAVS) and 32, 64 channel 3D LiDAR point clouds of the RELLIS-3D dataset collected in a real-world off-road environment. Our experimental results demonstrate that sensor and spatial domain shifts significantly impact the performance of LiDAR-based semantic segmentation models. In the absence of spatial domain changes between training and testing, models trained and tested on the same sensor type generally exhibited better performance. Moreover, higher-resolution sensors showed improved performance compared to those with lower-resolution ones. However, results varied when spatial domain changes were present. In some cases, the advantage of a sensor's higher resolution led to better performance both with and without sensor domain shifts. In other instances, the higher resolution resulted in overfitting within a specific domain, causing a lack of generalization capability and decreased performance when tested on data with different sensor configurations.

show abstract

Section: Lidar Datasetsmentioning

confidence: 99%

Analysis of LiDAR configurations on off-road semantic segmentation performance

Chen

Dabbiru

et al. 2023

Autonomous Systems: Sensors, Processing and Security for Ground, Air, Sea, and Space Vehicles and Infrastructure 2023

View full text Add to dashboard Cite

show abstract

“…The results of the potential traffic conflict analysis were utilized for road safety assessment, with special consideration given to CAVs based on insights provided by [45]. As is well known, potential traffic conflicts serve as the foundation for the application and modeling of the safety performance functions (SPFs), as elaborated in Section 5.…”

Section: Vissim Modelingmentioning

confidence: 99%

Connected Automated and Human-Driven Vehicle Mixed Traffic in Urban Freeway Interchanges: Safety Analysis and Design Assumptions

Granà,

Curto,

Petralia

et al. 2024

Vehicles

View full text Add to dashboard Cite

The introduction of connected automated vehicles (CAVs) on freeways raises significant challenges, particularly in interactions with human-driven vehicles, impacting traffic flow and safety. This study employs traffic microsimulation and surrogate safety assessment measures software to delve into CAV–human driver interactions, estimating potential conflicts. While previous research acknowledges that human drivers adjust their behavior when sharing the road with CAVs, the underlying reasons and the extent of associated risks are not fully understood yet. The study focuses on how CAV presence can diminish conflicts, employing surrogate safety measures and real-world mixed traffic data, and assesses the safety and performance of freeway interchange configurations in Italy and the US across diverse urban contexts. This research proposes tools for optimizing urban layouts to minimize conflicts in mixed traffic environments. Results reveal that adding auxiliary lanes enhances safety, particularly for CAVs and rear-end collisions. Along interchange ramps, an exclusive CAV stream performs similarly to human-driven ones in terms of longitudinal conflicts, but mixed traffic flows, consisting of both CAVs and human-driven vehicles, may result in more conflicts. Notably, when CAVs follow human-driven vehicles in near-identical conditions, more conflicts arise, emphasizing the complexity of CAV integration and the need for careful safety measures and roadway design considerations.

show abstract

“…In the traversability map, warmer colors indicate higher anticipated ease of traversal. systems in recent years, vision-based methods for traversability estimation have shown significant promise [4]- [8].…”

Section: > = >mentioning

confidence: 99%

“…The sparse labeling greatly reduces the tedium seen in strongly-supervised methods, while enabling easy explicit labeling of untraversable regions and not requiring robot experience during training. Our method can also provide continuous predictions of traversability (rather than discrete classes [8]), and does not require the additional assumptions, data, or models needed in selfsupervised methods [4]- [7].…”

Section: > = >mentioning

confidence: 99%

W-RIZZ: A Weakly-Supervised Framework for Relative Traversability Estimation in Mobile Robotics

Schreiber,

Sivakumar,

et al. 2024

IEEE Robot. Autom. Lett.

View full text Add to dashboard Cite

Successful deployment of mobile robots in unstructured domains requires an understanding of the environment and terrain to avoid hazardous areas, getting stuck, and colliding with obstacles. Traversability estimation-which predicts where in the environment a robot can travel-is one prominent approach that tackles this problem. Existing geometric methods may ignore important semantic considerations, while semantic segmentation approaches involve a tedious labeling process. Recent self-supervised methods reduce labeling tedium, but require additional data or models and tend to struggle to explicitly label untraversable areas. To address these limitations, we introduce a weakly-supervised method for relative traversability estimation. Our method involves manually annotating the relative traversability of a small number of point pairs, which significantly reduces labeling effort compared to traditional segmentation-based methods and avoids the limitations of selfsupervised methods. We further improve the performance of our method through a novel cross-image labeling strategy and loss function. We demonstrate the viability and performance of our method through deployment on a mobile robot in outdoor environments. Code is available at: https://github.com/ andreschreiber/W-RIZZ.

show abstract

CaT: CAVS Traversability Dataset for Off-Road Autonomous Driving

Cited by 26 publications

References 21 publications

Analysis of LiDAR configurations on off-road semantic segmentation performance

Analysis of LiDAR configurations on off-road semantic segmentation performance

Connected Automated and Human-Driven Vehicle Mixed Traffic in Urban Freeway Interchanges: Safety Analysis and Design Assumptions

W-RIZZ: A Weakly-Supervised Framework for Relative Traversability Estimation in Mobile Robotics

Contact Info

Product

Resources

About