IBISCape

IBISCape是一个专为多模态SLAM系统评估设计的模拟基准数据集，由巴黎-萨克雷大学和埃夫里大学的IBISC实验室创建。该数据集包含34个多模态数据集，适用于自主车辆导航，涵盖了从清晰到动态的不同天气和场景条件。数据集通过CARLA模拟器生成，利用Unreal Engine渲染高动态场景，提供包括立体RGB/DVS、深度、IMU和GPS在内的多种传感器数据，以及场景分割和车辆自我运动的真实数据。IBISCape不仅支持SLAM系统的性能评估，还解决了CARLA模拟中DVS和RGB摄像头的未知畸变参数问题，推动了动态和大规模环境下新多模态VISLAM技术的发展。

IBISCape is a simulated benchmark dataset specifically designed for evaluating multimodal SLAM systems, developed by the IBISC Laboratory from Paris-Saclay University and Évry University. This dataset includes 34 multimodal subsets tailored for autonomous vehicle navigation, covering diverse weather and scene conditions ranging from clear to dynamic environments. Generated via the CARLA simulator and rendered using Unreal Engine for high-dynamic scenarios, the dataset provides various sensor data including stereo RGB/DVS, depth, IMU, GPS, as well as ground-truth data for scene segmentation and vehicle egomotion. Beyond supporting performance evaluation of SLAM systems, IBISCape resolves the problem of unknown distortion parameters for DVS and RGB cameras in CARLA simulations, and promotes the advancement of novel multimodal VISLAM technologies in dynamic and large-scale environments.