Key components
Intelligent ingestion
Rule-based automatic ingestion and management capabilities enable scene indexing and post-processing of logs to enrich data with relevant information for testing and training of ML-based or integrated systems. Promptable, open-vocabulary perception foundation models process sensor data to automatically find events of interest.
Visualization
Ingest metrics from drive logs with built-in support for common formats (ROS bags, CAN, DDS, etc.). Play back drive logs in a cloud-based visualizer. Add logs as tiles and overlay sensors for accurate debugging.
Querying
Leverage structured queries and natural language to search across all metrics from recorded logs and enriched data, including object detections, behavior annotations, saved regions, and map entities. Use similarity search to rapidly find additional examples of a given edge case. Tag drive logs manually or with perception foundation models to surface relevant snippets quickly.
Curation and sharing
Collectively root-cause critical events and filter events automatically based on predefined rules, re-simulation results, or embeddings. Export raw sensor data or create simulation scenarios.
Key components
Intelligent ingestion
Rule-based automatic ingestion and management capabilities enable scene indexing and post-processing of logs to enrich data with relevant information for testing and training of ML-based or integrated systems. Promptable, open-vocabulary perception foundation models process sensor data to automatically find events of interest.
Visualization
Ingest metrics from drive logs with built-in support for common formats (ROS bags, CAN, DDS, etc.). Play back drive logs in a cloud-based visualizer. Add logs as tiles and overlay sensors for accurate debugging.
Querying
Leverage structured queries and natural language to search across all metrics from recorded logs and enriched data, including object detections, behavior annotations, saved regions, and map entities. Use similarity search to rapidly find additional examples of a given edge case. Tag drive logs manually or with perception foundation models to surface relevant snippets quickly.
Curation and sharing
Collectively root-cause critical events and filter events automatically based on predefined rules, re-simulation results, or embeddings. Export raw sensor data or create simulation scenarios.
.webp)
.webp)
