#HIROBA Documentation

Welcome to the HIROBA guides. HIROBA (広場 — "open square") is an AI-powered video conferencing platform built for studying hybrid human–AI social interaction. It combines real-time video, 3D AI avatars, speech synthesis, and a Wizard-of-Oz control surface inside one Zoom Video SDK–backed room.

These pages are aimed at three audiences. Pick the one that matches what you are trying to do.

#User Guide

For participants joining a HIROBA session.

• Joining a session from the lobby
• Using your mic, camera, and the in-call toolbar
• What the AI agents in the room do and how they decide to respond
• Leaving and rejoining

→ Open the User Guide

#Researcher Guide

For session hosts, study runners, and lab admins — the people designing experiments, configuring agents, and running sessions.

• Deploying HIROBA (Sakura VPS or AWS EC2 with Caddy + Let's Encrypt)
• Configuring conditions, agents, prompts, and avatars from the admin panel
• Running a session as host: the Session Host Panel, agent behaviour controls, the Wizard-of-Oz tab, participant management
• Collecting conversation data and exporting logs

→ Open the Researcher Guide

#Developer Guide

For people extending or modifying HIROBA's code. Architecture notes, code layout, the WebSocket sync protocol, the settings backend, and how to add new agent behaviours.

→ Open the Developer Guide

#Quick links

• Project on GitHub
• Hybrid Social Interaction Lab — University of Tsukuba

[!NOTE] HIROBA requires a secure context (https:// or localhost) for the camera and microphone to work — that is a browser-level requirement of getUserMedia, not something HIROBA imposes. In production deployments, Caddy provisions Let's Encrypt certificates automatically.