Back to Homepage

Emotion-Based Speed Control for Pac-Man

Driven by an ambition to advance human-computer interaction, this project integrates real-time affective computing with legacy gaming architecture. I engineered a sophisticated control loop that utilises TensorFlow Lite for facial emotion recognition, dynamically modulating the game's difficulty in response to the player's emotional state. The result is a bio-responsive gaming experience that creates a novel, immersive feedback loop between user sentiment and digital environment.

Project VIII

60Hz Sampling Rate
7-Class Classifier
~14ms Latency
Bio-Feedback Loop

Technical Stack

Python TensorFlow Lite OpenCV WebSockets JavaScript

Advanced Affective Computing Interface

Developed a sophisticated system capable of discerning a range of user emotions in real-time, utilising a pre-trained TensorFlow Lite model that processes live video feeds to accurately assess facial expressions and affective states, providing valuable input for dynamic game adjustments.

Dynamic Gameplay Adaptation

Engineered a highly responsive and adaptive control system that dynamically modifies the pace of the Pac-Man game, reacting to the nuances of emotional states detected through a bespoke socket communication protocol, resulting in a more personalised and immersive gaming experience.

Modular and Scalable Architecture

Constructed a fully modular system architecture for seamless integration of all its component parts (emotion recognition, video frame analysis, and in-game control), whilst also guaranteeing scalability and facilitating rapid development cycles and future upgrades.