Foxglove

Foxglove is a purpose-built observability platform for robotics teams building Physical AI systems. It provides comprehensive tools for collecting, analyzing, and learning from the vast quantities of multimodal data required to build, train, deploy, and operate reliable robots.

The platform supports the full robotics development lifecycle from prototyping to production. Engineers can visualize live robot data or playback recorded sessions using over 20 customizable panels including 3D scenes, image viewers, plotting tools, maps, and diagnostic displays. Data formats supported include ROS 1 bags, ROS 2 MCAP and DB3 files, PX4 ULog, and custom formats via extensible data loaders.

Live connections are established through the Foxglove WebSocket bridge, Rosbridge, or native ROS 1 connections. The Foxglove Agent simplifies data ingestion from edge devices, while the cloud platform enables team collaboration through shared layouts, event annotations, and centralized data management.

Key capabilities

Foxglove provides visualization panels for 3D scenes with point clouds and markers, camera image feeds with bounding box overlays, time-series plotting, geographic mapping for mobile robots, and raw message inspection. The layout system allows teams to create reusable visualization configurations tailored to specific debugging scenarios.

The data management platform indexes recordings by device, time, and topic with configurable retention policies. Webhooks enable integration with existing data pipelines and CI/CD systems. The platform supports both cloud-hosted and self-hosted deployment options for enterprises with specific compliance requirements.

Limitations

• No native OPC-UA or industrial protocol support; focused specifically on ROS-based robotics stacks
• Actions (ROS actions) are not supported in any live connection mode
• Desktop app required for native ROS 1 connections; web app limited to WebSocket connections
• Free tier limited to 3 users and 10GB cloud storage
• Self-hosted deployment only available on Enterprise plan
• MCAP format adoption still growing; legacy ROS 2 DB3 files lack embedded message definitions