Overview

This project won the Best Robotic Project Prize at Imperial College for the second year cohort. The autonomous campus tour guide robot integrates advanced robotics, computer vision, and cloud technologies to provide interactive campus tours.

Key Features

• Self-balancing Mobility: Dual-loop PID/LQR control achieving <15cm positional accuracy and 3s disturbance recovery
• Reinforcement Learning: Developed TD3 reinforcement learning in Gazebo with BARN dataset for trajectory following
• Computer Vision: YOLOv8 Nano vision system integrated on RK3588 achieving >90% person-tracking accuracy at <100ms latency
• Cross-platform Control: Flutter app for robot control with real-time weather sensing (GY-39/GPS)
• Cloud Infrastructure: Deployed AWS EC2 cloud infrastructure with JWT-secured REST API, PostgreSQL, and ReactJS dashboard
• Voice Assistant: Implemented RAG-enhanced voice assistant using OpenAI API with on-device MP3 playback via ESP32 I2S