Rbyte: Multimodal Spatial Intelligence and Robotics Data in MCAP Format

Rbyte provides multimodal datasets for spatial intelligence and robotics, released by yaak-ai and updated in February 2026. The collection utilizes MCAP and TensorDict formats to facilitate high-performance spatial computing and integration with PyTorch and Polars.

Use Cases

• Training spatial intelligence models using the TensorDict integration for state-action mapping
• Visualizing robotics sensor streams through the Rerun logging framework
• Analyzing high-frequency spatial logs stored in the MCAP format

Strengths

• Apache-2.0 licensed for open research and commercial use
• Native support for MCAP and TensorDict formats for robotics workflows
• Integration with Polars for high-speed data manipulation

Limitations

• Specific record counts and sensor modalities are not publicly documented
• Requires specialized tooling such as Rerun for data visualization

Provenance

Source

yaak-ai

Collection Method

sensor

Freshness

Updated as of February 2026.