Explore MCAP benefits in the latest Isaac ROS release
In robotics and embodied AI, data logging and analysis play pivotal roles in training, evaluating, and debugging system performance. As the complexity of robotics systems escalates, the need for an efficient, robust, and scalable logging format becomes paramount. That is why we developed MCAP – an open-source container format for multimodal logging.
This week, NVIDIA announced Isaac ROS 3.0, which includes a new data recorder package for multi-sensor data capture to ROSBag using MCAP. Isaac ROS is a collection of hardware-accelerated, high-performance, low-latency ROS 2 packages for building autonomous robots that leverage the power of Jetson, Nova, and other NVIDIA platforms.
NVIDIA is our latest partner to adopt and recommend MCAP. Since MCAP launched in 2022, it has rapidly gained popularity in the robotics industry because of its reliability and performance benefits. It is the preferred logging format for some of the largest commercial robotics companies in the world, and is also the ROS 2 default storage recording format.
Traditional logging formats often struggle with multimodal data, from sensor readings and video feeds to robot pose and actions. MCAP simplifies this complexity by providing a unified container format, facilitating easier data handling, storage, and retrieval, enabling you to focus more on training and analysis, and less on managing data.
MCAP supports pluggable message serialization, with optional compression ensuring rapid read and write operations even with large datasets. This efficiency is crucial for real-time systems where delays can compromise system integrity and performance, especially as data volumes grow with scale.
MCAP files are also completely self-contained, meaning message schemas are embedded in the file, and no additional dependencies are required for decoding. This promotes interoperability so teams can easily exchange data without needing custom conversion tools.
If you’re currently using a bespoke format for storing multimodal data, MCAP is a great alternative. We provide libraries in C++, Python, Go, Swift, Rust, and TypeScript, as well as a suite of conformance tests to facilitate interoperability with other tools. We also support opening and visualizing MCAP files in Foxglove to easily visualize the data in a customizable dashboard.
Get started with the developer documentation at https://mcap.dev/ and all your data visualization needs using Foxglove.
