Interactive Game Hub -

AI-Driven Games

The
Project

Interactive Game Hub is an experimental system that explores how large language models, gesture recognition, and everyday objects can become playful interfaces for digital experiences. Instead of relying on traditional controllers, the project invites players to communicate with the computer through their bodies and their surroundings — using hand gestures, physical objects, and webcam‑based recognition as the primary modes of interaction.

The result is a small hub containing two games that each test a different interaction paradigm: Cat Escape, controlled entirely through gestures, and Riddle Object Hunter, driven by object recognition and LLM‑generated hints.

Project Type: AI-driven Browser Game
Project Format: Group project
Role: Game Designer, Programmer
Tools: Cursor, HTML, CSS, JavaScript, LM Studio Vision.

Interaction Approach

Interactive Game Hub is an experimental system that explores how gestures, everyday objects, and LLM‑driven interpretation can become playful alternatives to traditional game controls. The project combines computer vision, hand‑tracking, and object recognition to let players interact with two different games through their bodies and their surroundings. In Cat Escape, movement and interaction are controlled entirely through gestures: the player shows their palm to activate tracking, guides the cat with their index finger, and collects keys or opens the door through hand poses. In Riddle Object Hunter, the interaction shifts to object‑based play, where an LLM provides riddles and interprets webcam‑captured objects to determine whether the player has found the correct item. Together, these mechanics demonstrate how multimodal inputs and AI reasoning can create more embodied, intuitive, and exploratory forms of gameplay.

Game 1 - Cat Escape

Cat Escape is a gesture‑controlled game where the player helps a cat escape a locked room. The objective is to collect five keys scattered across the field while avoiding bombs. The player begins by showing their open palm to the webcam, allowing the system to detect and track their hand. Movement is then controlled through the index finger, which sets the direction of the cat’s motion. To interact with objects — keys or the final door — the player shows their palm again, triggering the system to “pick up” items. The game gives the player three lives, allowing for mistakes while still maintaining a sense of challenge. The entire experience is built around bodily interaction: the player navigates, collects, and avoids obstacles purely through gesture‑based input.

Game 2 - Riddle Object Hunter

Riddle Object Hunter shifts the interaction from gestures to object recognition and LLM‑driven reasoning. The player’s goal is to wake up a spirit by presenting the correct object to the webcam three times. Each round begins with a riddle generated or interpreted by the LLM, offering hints about the object the player must find. The player then shows different objects to the camera. The system analyzes the image, compares it to the riddle’s clues, and responds through the spirit’s reactions. If the object is incorrect, the spirit reacts negatively; if it matches the riddle, the spirit acknowledges it and progresses to the next clue. Completing three riddles successfully wins the game. This game explores how AI can interpret ambiguous, real‑world inputs and turn them into playful, meaning‑making interactions.