Skip to content

FAQ

This section aims to answer some of the common questions users may encounter while using the dataset.

How can I obtain the dataset?

The dataset can be obtained by contacting Realman's official channels.

What format is the dataset in?

The dataset primarily uses the MCAP (Modular Containerized Autonomous Data) format. This is an open-source container format optimized for robotic applications, supporting efficient storage, fast access, and multimodal data.

How can I read and parse MCAP files?

You can use various tools and libraries to read MCAP files, such as Foxglove and third-party data viewing software. Foxglove Studio is a powerful visualization tool that can directly open and replay MCAP files. The ros2 bag tool in ROS 2 also supports the MCAP format.

What types of sensor data are included in the dataset?

The dataset contains rich multimodal data, mainly including RGB images and depth images from the head and arms, proprioceptive data (joint positions, angles, velocities, acceleration), and fine tactile data from the end-effectors. In addition, it may also include some electrical data such as current, voltage, and temperature information from the joints.

Is the data annotated? What is included in the annotations?

Annotations can be provided based on user requirements. Annotation content includes, but is not limited to, task-level information (task ID, description, success/failure status), action-level information (action sequence segmentation, start/end timestamps, action type labels, keyframes), object-level information (object detection and tracking, object state changes), and robot state annotations (key state changes, abnormal events).

Can I use the dataset for commercial purposes?

The use of the data must comply with the data usage license agreement. Generally, publicly released datasets are mainly intended for academic research or non-commercial purposes. If you have a need for commercial use, please contact us through the Developer Center to discuss the possibility of collaboration.

What should I do if I encounter problems or find errors while using the dataset?

You are welcome to contact us through the feedback channels provided by the Developer Center (such as forums, emails). We will reply and handle your issues as soon as possible.

Will the dataset be updated or expanded in the future?

Yes, we plan to regularly update and expand the dataset based on user feedback and research progress, adding new tasks, scenarios, or sensor data. Please stay tuned to the notifications on the Developer Center.