BantayMuseo

Source repository is private; public source code is planned after the World Robot Olympiad.

Overview

BantayMuseo is a compact, portable semi-autonomous social robot designed to bridge the gap between rural Filipino students and Philippine cultural heritage. Inspired by the Philippine jeepney, its physical design houses a tablet interface and onboard computing system that provide visual storytelling, audio narration, voice interaction, interactive cultural learning, and AI-assisted exploration. The robot can operate both as a supplementary museum guide and as a mobile cultural education platform that can be transported to rural and underserved communities.

At its core, BantayMuseo uses a Raspberry Pi as its central processing unit, hosting the Express and FastAPI backend services, curated cultural knowledge base, offline AI system, and communication services. A locally hosted Llama 3.2 1B model provides conversational cultural assistance without requiring an internet connection. A lightweight Retrieval-Augmented Generation (RAG) system retrieves relevant information from the curated cultural database before generating responses, helping constrain the AI to trusted Philippine cultural content and reduce hallucinations.

For physical navigation, a LEGO SPIKE Prime Hub controls the robot's tracked drivetrain while wheel encoders and a color sensor provide movement and location feedback. When a visitor selects a gallery or destination, the Raspberry Pi communicates with the SPIKE Prime Hub through Bluetooth Low Energy using Bleak. The robot combines encoder-based distance tracking with floor markers and exhibit-zone detection to autonomously navigate between designated areas. The modular architecture also allows the navigation subsystem to be migrated to lower-cost hardware in future versions.

BantayMuseo (Bantay Museo) was developed in collaboration with the National Museum of the Philippines - Bohol and is intended as a supplementary educational tool rather than a replacement for human museum docents. It was built by the Ubay National Science High School (UNSHS) Future Innovators team: Elijah Juris Evangelista (Juris) as team lead, programmer, and 3D designer, alongside teammates Jianna Marie Nemenzo, Rimar Benedict Millanes, Alyna Sumampong, and Perkz Johann Ong. Representing Bohol, the team won the Future Innovators Category at the 25th Philippine Robotics Olympiad and is now preparing to represent the Philippines in the Future Innovators Junior Category at the World Robot Olympiad (WRO) 2026 in Puerto Rico, under this year's "Robots Meet Culture" theme. Its primary innovation is portability: instead of requiring rural students to travel long distances to access cultural institutions, the robot can bring interactive heritage education directly into classrooms and underserved communities. By combining affordable robotics, offline AI, localized cultural knowledge, and culturally inspired design, BantayMuseo aims to make Philippine history and heritage more accessible, engaging, and resilient to unreliable internet connectivity.

My Role

Main developer, Team Lead, Programmer, and 3D Designer/Fabricator. I designed and implemented the system architecture, frontend, backend, AI integration, and robot control components. I also designed custom 3D components for the robot and fabricated them using 3D printing.

Structure

• front — Interactive museum galleries, exhibit media viewers, AI chat, voice interaction, quiz, music player, and navigation controls.
• back — Node.js and Express API layer that proxies AI, speech transcription, and robot-control requests.
• ai — Local Ollama language model using Llama 3.2 1B with museum-specific cultural context and locally stored knowledge.
• robot — Python and FastAPI services communicating with a LEGO Pybricks hub over Bluetooth Low Energy.
• navigation — LEGO SPIKE Prime-based robot navigation using wheel encoders and color sensors to detect movement and exhibit zones.
• offline — Service worker caches museum pages, scripts, media, and exhibit assets so core museum content remains available without internet access.

Technologies

Results & limitations

• The project is still under active development.
• AI responses are limited by the locally available cultural knowledge and model capabilities.
• Offline AI performance is constrained by the available Raspberry Pi hardware.
• Robot functionality depends on the connected LEGO/Pybricks hardware and Bluetooth communication.
• The frontend source in the development structure requires the appropriate deployment/build step before being served by the backend.