ECMD: An Event-Centric Multisensory Driving Dataset for SLAM

TL;DR

ECMD is a comprehensive multisensory dataset for autonomous driving, combining diverse sensors and challenging scenarios to advance SLAM research and improve localization accuracy in complex environments.

Contribution

The paper introduces ECMD, a new multisensory dataset with synchronized sensors and challenging driving scenarios, enabling better SLAM algorithm benchmarking and development.

Findings

Evaluated state-of-the-art SLAM algorithms on ECMD

Identified limitations of current SLAM methods in challenging scenarios

Provided high-precision ground truth for benchmarking

Abstract

Leveraging multiple sensors enhances complex environmental perception and increases resilience to varying luminance conditions and high-speed motion patterns, achieving precise localization and mapping. This paper proposes, ECMD, an event-centric multisensory dataset containing 81 sequences and covering over 200 km of various challenging driving scenarios including high-speed motion, repetitive scenarios, dynamic objects, etc. ECMD provides data from two sets of stereo event cameras with different resolutions (640*480, 346*260), stereo industrial cameras, an infrared camera, a top-installed mechanical LiDAR with two slanted LiDARs, two consumer-level GNSS receivers, and an onboard IMU. Meanwhile, the ground-truth of the vehicle was obtained using a centimeter-level high-accuracy GNSS-RTK/INS navigation system. All sensors are well-calibrated and temporally synchronized at the hardware level, with recording data simultaneously. We additionally evaluate several state-of-the-art SLAM algorithms for benchmarking visual and LiDAR SLAM and identifying their limitations. The dataset is available at https://arclab-hku.github.io/ecmd/.