Camera Configuration
Configure on-robot cameras so operators can stream live video during teleoperation.
The agent discovers V4L2 cameras on the robot's Linux host and streams them to the operator during a teleop session. Video is encoded on the robot and delivered over a real-time peer connection to the operator's browser.
Requirements
- Linux host with the camera(s) exposed as V4L2 devices.
- One or more cameras supported by V4L2.
What the agent handles
The agent discovers V4L2 devices, encodes frames for low-latency delivery, supports dual-track streaming, and live-switches feeds during a session. For the full streaming pipeline, supported input formats, and dual-track/live-switching details, see the agent's Cameras reference.
Selecting cameras
During a session the operator can request a specific camera (or pair) for the live tracks. The agent switches feeds without dropping the session. See Cameras for how selection and live switching work.
Configuration
Detailed camera configuration options are coming soon: where selection and preferences
are set, how to pin specific /dev/video* devices, and how to set resolution, format,
and framerate.
Tip. The agent prioritizes latency over quality for teleop. Prefer camera modes that keep end-to-end latency low.