CREATE数据库

CREATE数据库是由舍布鲁克大学电气与计算机工程系创建的，包含14小时的多模态记录数据，来源于基于iRobot Create®的移动机器人平台。该数据集涵盖视觉、听觉、电机和本体感受等多种传感器数据，旨在支持移动机器人在室内环境中学习多模态环境表示。数据集内容丰富，包括RGB摄像头、光学流场、麦克风、惯性测量单元等数据，数据量庞大，Tokens数众多。创建过程中，使用了标准化的消息和主题，确保与ROS社区的兼容性。该数据集适用于多种研究领域，如无监督对象学习、多模态预测、自我运动检测和因果关系检测，旨在解决机器人在复杂环境中的自主导航和交互问题。

The CREATE Database was developed by the Department of Electrical and Computer Engineering at Université de Sherbrooke. It comprises 14 hours of multimodal recorded data collected from a mobile robotic platform based on the iRobot Create®. This dataset covers diverse sensor data including visual, auditory, motor, and proprioceptive modalities, and is designed to support mobile robots in learning multimodal environmental representations in indoor environments. The dataset is rich in content, including data from RGB cameras, optical flow fields, microphones, inertial measurement units (IMUs), etc., with a large volume of data and a substantial number of tokens. During its development, standardized messages and topics were utilized to ensure compatibility with the ROS (Robot Operating System) community. This dataset is applicable to a wide range of research fields, including unsupervised object learning, multimodal prediction, ego-motion detection, and causal relationship detection, aiming to address the challenges of autonomous navigation and interaction of robots in complex environments.