The ECMD Datasets

This is first event-based SLAM datasets specifically focus on urbanized autonomou driving. We explore the inquiry: Are event cameras ready for autonomous driving? At the same time, we also investigates the perceptual capabilities of various sensors, including LiDAR, standard cameras, infrared cameras, and GNSS-RTK/INS.
The contributions of our work can be summarized as follows:

• Our sensor platform consists of various novel sensors, including two sets of stereo event cameras with distinct resolutions (640×480, 346×260), an infrared camera, stereo industrial cameras, three mechanical LiDARs (including two slanted LiDARs), an onboard inertial measurement unit (IMU), and two global navigation satellite system (GNSS) receivers. For the ground-truth, we adopt a centimeter-level position system that combines the GNSS real-time kinematic (RTK) with the fiber optics gyroscope integrated inertial system as GNSS-RTK/INS.
• ECMD collects 81 sequences covering over 200 kilometers of trajectories in various driving scenarios, including dense streets, urban, tunnels, highways, bridges, and suburbs. These sequences are recorded under daylight and nighttime, providing challenging situations for Visual and LiDAR SLAM, e.g., dynamic objects, high-speed motion, repetitive scenarios, and HDR scenes. Meanwhile, we evaluate existing state-of-the-art visual and LiDAR SLAM algorithms with various sensor modalities on our datasets. Moreover, our dataset and benchmark results are released publicly available on our website.

We hope that we can make some contributions for the development of event-based vision, especially event-based multi-sensor fusioin for autonomous driving.
The visualization of each sequence is available in Download section and Bilibili.
If you have any suggestions or questions, do not hesitate to propose an issue to our Github Repository.