<?xml version="1.0" encoding="UTF-8"?><rss version="2.0"><channel><title>OpenHRI research</title><description>OpenHRI builds open benchmarks, models and tools for conversational agents that listen, look and take turns in real time.</description><link>https://openhri.ai/</link><item><title>Two Faces of a Turn Exchange</title><link>https://openhri.ai/research/turn-taking-faces/</link><guid isPermaLink="true">https://openhri.ai/research/turn-taking-faces/</guid><description>What the listener&apos;s and the speaker&apos;s faces show around the moment the floor changes hands, measured in 2,670 conversations filmed in the same room and over video calls.</description><pubDate>Sat, 14 Nov 2026 00:00:00 GMT</pubDate></item><item><title>Listening through the pause: streaming end-of-turn detection</title><link>https://openhri.ai/research/smart-turn-streaming/</link><guid isPermaLink="true">https://openhri.ai/research/smart-turn-streaming/</guid><description>Stock Smart Turn asks its model once, 0.2 s after you stop talking. Asking every 100 ms with a model that knows how long the silence has lasted cuts false turn ends on real meetings from 42 % to 5 %.</description><pubDate>Sun, 25 Oct 2026 00:00:00 GMT</pubDate></item><item><title>Do video-call agents look? An open reproduction of VideoFDB</title><link>https://openhri.ai/research/open-videofdb/</link><guid isPermaLink="true">https://openhri.ai/research/open-videofdb/</guid><description>We rebuilt NVIDIA&apos;s full-duplex audio-visual benchmark as runnable code, reproduced its human reference, and scored two realtime APIs and three avatar products. The agents still do not use the camera.</description><pubDate>Fri, 16 Oct 2026 00:00:00 GMT</pubDate></item></channel></rss>