Mobile Robot
ROS packages coordinate robot bringup, TF, motors, lidar, web connection, and the display workflow.
TL;DR: Hongdo is an HRI-focused promotional robot that moves through event spaces, approaches detected visitors, speaks with them, and offers a traditional Korean-style web experience that transforms participant photos into drawings.
The system combines ROS bringup, motor control, perception, obstacle sensing, a browser-based robot display, and AI image conversion. Hongdo was deployed at four public festivals, starting with Ansan Everyone's Playground Festival (안산 모두의 놀이터 축제); this page highlights the second deployment, Ansan Youth Day Festival, where visitors interacted with the robot through the event UI.
System Overview
Visitor Interaction
Human detection and obstacle modules support event-space approach behavior and safe movement.
AI Photo Experience
The web interface guides photo capture, style selection, image conversion, and result sharing.
Architecture
Hongdo is organized as a ROS workspace. The NUC-side packages handle bringup, web integration, teleoperation, TF broadcasting, obstacle sensing, and robot description. The web layer is served from Node.js and connects the robot display to the ROS workflow.
- hongdo_ros_bringup starts robot, model, motor, and web-connect launch flows.
- hongdo_ros_web serves the event interface and photo conversion flow.
- hongdo_ros_obstacle integrates lidar and obstacle messages.
- hongdo_ros_description stores URDF and mesh assets for the robot model.
Robot Display UI
The on-robot interface turns the technical pipeline into a guided visitor experience. Its ink-wash landscape, illustrated mascot, and short prompts keep the interaction approachable in a busy festival setting.
-
Welcome & Consent
Hongdo introduces the experience and asks whether the visitor wants to participate.
-
Photo Capture
The camera workflow guides the visitor through taking and confirming a portrait.
-
Style & Conversion
The visitor selects an effect while the interface carries the image through conversion.
-
Result & Sharing
The finished artwork is displayed and made available through a QR-based handoff.
Photo-To-Art Pipeline
The conversion flow references open-source image translation and sketch-generation projects, then wraps those outputs in a robot-facing event interaction. The visitor sees a continuous experience: capture, select, convert, and receive.
Festival Deployments
Hongdo was deployed at four public festivals. The records below are organized by deployment so each festival can carry its own media, venue context, and field notes.
Festival 01
Ansan Everyone's Playground Festival (안산 모두의 놀이터 축제)
Hongdo's first public deployment introduced the mascot robot form and a camera-linked booth workflow.
- Robot mobility was not yet enabled.
- Camera integration was connected.
Festival 02
Ansan Youth Day Festival
The second deployment added navigation behavior for live visitor-space interaction.
- Obstacle avoidance was added.
- Human following was added.
Festival 03
Ansan Citizens' Day Festival (안산시 시민의 날 축제)
The third deployment extended the public interaction workflow with AI output and physical expression.
- AI photo-conversion version was added.
- Robot joint motion was added.
Festival 04
Hongdo Festival (홍이와 먹이 축제)
The fourth deployment validated the complete system in a dedicated public festival environment.
- AI photo-conversion version was added.
- All systems operated normally.
- Tested with more than 100 citizens.
Behind the Scenes
Fabrication and field debugging are collected separately from the festival-by-festival deployment record.
Finishing the robot head shell before deployment.
Fabrication workspace with the head shell and face parts in progress.
