πŸ€– Tsinghua AI Challenge:Real-Time Strategy and Space Simulation

Mar 20, 2024Β·
Rongkui Zhang
Β· 2 min read

Overview

This project showcases the development of intelligent algorithms for controlling spacecraft in a real-time strategy simulation during the Tsinghua AI Challenge. The system integrates resource management, construction, and combat mechanics within competitive constraints, earning the Third Prize (Top 10%).

ai in Action

Competition Highlights

  • Gameplay Overview:

    • Two teams compete in each match, played on four distinct maps.
    • Objectives include constructing buildings, extracting resources, and destroying enemy bases.
    • Each match ends when either:
      • One team’s base is destroyed.
      • The match duration (10 minutes) elapses.
    • Scores are calculated after each match, and the team with the higher score wins.

  • Task Structure:

    • Players control five characters (PlayerID 0–4), including:
      • 1 Base (PlayerID 0): Acts as the team’s headquarters.
      • 4 Ships (PlayerID 1–4): Divided into:
        • Civil Ships: Maximum 2 per player.
        • Military Ships: Maximum 2 per player.
        • Flagship: Maximum 1 per player.

Project Achievements

  • Developed real-time algorithms in C++ for:
    • Optimizing resource extraction from predefined points using AI-controlled paths.
    • Managing the construction of buildings at strategic locations.
    • Directing ships during combat using decision-making systems.
  • Designed an adaptive navigation system with an A* pathfinding algorithm for optimal movement.
  • Implemented combat strategies that leverage energy, positioning, and multi-agent coordination to eliminate enemies and protect assets.

Technical Architecture

State Machine Flow

graph TD A[Start] -->|Identify Objective| B[Resource Collection] A -->|Defend or Attack| C[Combat Management] B --> D[Construction Planning] D --> E[Base Protection] C --> E E -->|Victory Check| F[Match End]

Key Features:

  1. A Pathfinding*: Facilitates efficient movement across complex maps.
  2. Modular Communication:
    • Implemented UART-based protocols for seamless real-time communication between agents.
  3. Multi-Agent Coordination:
    • Integrated algorithms for dynamic role switching between offensive, defensive, and resource collection tasks.

Challenges and Solutions

  1. Resource Scarcity:
    • Balanced team strategies to maximize resource extraction efficiency.
    • Addressed potential contention through optimized task allocation.
  2. Navigation Obstacles:
    • Enhanced pathfinding accuracy using map preprocessing and heuristic adjustments.

Thank you for exploring this project. Feel free to share your thoughts! πŸ™Œ

AI Challenge Real-Time Strategy Space Simulation Embedded Systems

Related