Embodied AI Glossary中文

Web Real-Time Communication (WebRTC)

WebRTC 实时音视频传输WebRTCAdvanced

A browser-native standard for low-latency, peer-to-peer audio, video, and data transfer.

WebRTC is an open standard jointly developed by the W3C (web APIs) and the IETF (network protocols), driven primarily by Google, and it became an official W3C recommendation in 2021. It lets browsers and applications send audio, video, and arbitrary data directly to each other, peer to peer, without routing through a server, with built-in codecs, jitter buffering, and congestion control; end-to-end latency can typically be kept under a few hundred milliseconds. The two ends are often on different local networks, so an ICE process working with STUN/TURN servers is needed to get through NAT (the address translation done by routers). In embodied AI, it's commonly used to stream a robot's camera feed live to a remote operator, and some robot makers also use it to connect a phone app to the robot.

ExampleDuring remote teleoperation, the robot encodes its head-camera feed and pushes it over WebRTC to the operator's browser or VR headset, with control commands sent back over WebRTC's data channel.

Also called
WebRTC
Related
WebSocket · Teleoperation · Control Latency · Vuer · FFmpeg
Sources
WebRTC 官网 (Chinese)
WebRTC: Real-Time Communication in Browsers (W3C)

See it in the full glossary →