Publication of the LUCID Dataset

Last updated on Dec 11, 2022 1 min read

We are excited to announce the release of the LUCID Dataset, a novel synthetic street-view LiDAR dataset designed to advance computer vision and object detection in the automated driving domain. Comprising 100 scenarios with approximately 1.1 hours of drive time, 40,000 frames, and 500,000 bounding boxes, the LUCID Dataset is a significant resource for researchers and developers.

The LUCID dataset has been generated using Eclipse MOSAIC alongside with SUMO for route and traffic density modeling, and CARLA for simulating advanced LiDAR sensors and environments, following an extensive analysis of other datasets like KITTI, nuScenes and PreSIL. This innovative approach allows for effective cross-dataset generalization, bridging the gap between real and synthetic LiDAR data.

For a comprehensive overview of our methodology, we invite you to read our paper titled “Cross-Dataset Generalization: Bridging the Gap between Real and Synthetic LiDAR Data” which we presented at the SIMUtools 2024 Conference (taking place 09-10 December 2024).

The LUCID Dataset is free for download. It is organized in a familiar structure, following KITTI formatting, and includes both labeled and raw data. To facilitate initial exploration, we provide four sample scenarios, with the full dataset available for download.

For additional information and download links, please visit our dedicated website: https://www.dcaiti.com/research/simulation/lucid

Related

featured.png
AI-NET-ANTILLAS - Perception for Remote Operated Driving

After four years of research, the AI-NET-ANTILLAS project has concluded, and the final event took place in conjunction with the Berlin 6G Conference 2024. Collaborating with our partners, we integrated several key components: Cloud-based LIDAR processing, next generation networks, and artificial intelligence, to create a new service and application platform.

» Read Post
featured.png
Simulated Testing of Traffic State Estimation using Eclipse MOSAIC

Daily commutes can quickly turn into stressful affairs when overcrowded roads become congested and what would be a 20 minute ride becomes an hour of staring at the license plate of your fellow comrade-in-traffic. Real-time Traffic State Estimation (TSE) aims to alleviate this strain by predicatively recognizing congested areas and offering alternative routing options. In a recent publication we established a framework for easy implementation and evaluation of novel TSE systems.

» Read Post
featured.png
Introducing the BeST Scenario

We present a large-scale traffic scenario of Berlin for the Evaluation of Smart Mobility Applications. This scenario provides 24 hours of motorized private traffic, including almost 2,25 million individual trips. The Berlin SUMO Traffic (BeST) scenario is open-source and prepared to be compatible with Eclipse MOSAIC.

» Read Post