AI-Powered Puzzle & Racing Game for Elderly Care Center (Unity 3D+ IoT + ML AI model)

Role: Project Lead | Duration: ~12 months | Technologies: Unity, C++, Python, IoT, AI Pose Recognition | Team members: 7

📌 Overview

A final-year undergraduate project delivering an AI-driven, IoT-integrated puzzle & racing game tailored for an elderly care center.

The system integrates real-time pose recognition AI with Unity gameplay, enabling intuitive motion-based controls via IoT sensors to boost physical activity, cognitive engagement, and rehabilitation outcomes.

💡 My Role & Responsibilities

• End-to-End Leadership: Oversaw concept creation, client engagement, system architecture, development, testing, and final delivery.
• Game design: design game content and story background catering to Singapore culture, adjusting difficulty as per the elderly’s cognitive and physical capabilities.
• AI Integration: Built and trained a custom Human Activity Recognition ML model (~90% accuracy) using self-collected motion data; integrated into Unity for real-time interaction.
• Gameplay Programming: Developed core racing mechanics in C# with physics-based challenges adapted for elderly accessibility.
• Dev details & Asset Management:
  • Outsourced and managed 3D art and UI asset creation to external designers due to a lack of in-house art expertise, ensuring visual quality and consistency with game design.
  • Integrated 2D sprite animations using Unity Animator.
  • Use Spine to control the character's motion.
  • UI/UX Design: Create a leaderboard and score system using the Unity Canvas system.
  • Use baking to reduce loading time.
• IoT Hardware Integration: Connected ESP32 sensors to Unity via WiFi UDP communication for real-time pose feedback.
• Team Management: Led a 7-member team in Agile sprints, conducted weekly code reviews, and ensured milestone delivery.

⚙️ Tech Stack

• Game Engine: Unity (C#) + Python modules
• AI/ML: Keras, Scikit-learn, LSTM, autoencoder