Elite Snake AI - Tournament Championship Edition

An intelligent multiplayer Snake AI designed for competitive tournament play

Quick Start Top Agents AI Strategy Setup

---

Overview

Elite Snake AI agent using A pathfinding*, Hamilton cycles, 3-step opponent prediction, and territory control to dominate competitive tournaments.

Performance

• Survival Rate: 85%
• Decision Speed: <30ms (optimized for 50ms intervals)
• Hunt Success: 70%+ against smaller snakes

For Non-Programmers

Think of it like a video game where computers control snakes. Each snake tries to:

1. Eat apples to grow
2. Trap other snakes
3. Avoid collisions
4. Become longest to win

Every 50ms (0.05 seconds), the AI calculates: Up, Down, Left, or Right - like chess in real-time!

---

Quick Start

# Clone repository
git clone https://github.com/SindyMl/Snake-AI.git
cd Snake-AI

# Development mode (single snake with visualization)
java -cp "src;lib/SnakeRunner.jar" MyAgent -develop

# Tournament mode (4 snakes)
java -jar lib/SnakeRunner.jar -j submissions/Snake-AI.jar -j submissions/beatMedium.jar -j submissions/MyAgent.jar -j submissions/sub3.jar

# Build your own
.\build.ps1

---

Project Structure

Snake-AI/
 lib/
    SnakeRunner.jar              # Game engine
 src/
    MyAgent.java                 #  Main AI (Snake-AI.jar source)
 submissions/
    Snake-AI.jar                 #  #1 Elite Tournament Edition
    beatMedium.jar               #  #2 Aggressive Hunter
    MyAgent.jar                  #  #3 Balanced Strategy
 images/
    local.png
    Tournament.png
 documentation/                    # Technical docs
 build.ps1                         # Build script
 snake_config.txt                  # Game config

---

Top Performing Agents

1st Place: MyAgent.java

Source: src/MyAgent.java submissions/Snake-AI.jar

Stats:

• 85% survival
• 3.2 avg kills/game
• Max length: 127 segments

Strengths:

• 3-step trap prediction
• Territory control (5-tile radius)
• Adaptive aggression
• 50ms time budget optimization

2nd: beatMedium.jar

• 78% survival, 2.8 avg kills, max 105 segments
• Fast BFS, strong defense

3rd: Snake_AI.jar

• 72% survival, 2.1 avg kills, max 92 segments
• Reliable space management

---

How to Export JAR in VS Code

Step-by-Step

1. Open VS Code with Snake-AI project

2. Find "JAVA PROJECTS" panel (left sidebar, bottom)

   • If missing: Ctrl+Shift+P "Java: Configure Java Runtime"

3. Click Export Jar button ( icon)

4. Select Options:

   • "Export as executable JAR"
   • Select all files
   • Save to: submissions/MyAgent_[date].jar

5. Test:

   java -jar submissions/MyAgent_[date].jar

Alternative: Build Script

.\build.ps1
# Auto-compiles and creates JAR in submissions/

---

AI Strategy & Methods

Complete Feature List

Feature	Description
A Pathfinding*	Opponent-aware optimal paths (+3 to +8 enemy penalties)
Multi-Step Prediction	3-turn opponent movement forecast
Territory Control	5-tile radius dominance (+25/tile bonus)
Hamilton Cycle	Zigzag space-filling pattern (guaranteed safe)
Flood-Fill	Reachable space calculation (cached per turn)
Escape Route Blocking	Trap mechanics (+300/blocked exit)
Dynamic Apple Logic	Value-based pursuit/avoidance
Adaptive Aggression	Hunt when strong, evade when weak
Emergency Escape	Zigzag/circle evasion patterns
Space Caching	50% performance boost

Core Algorithms

1. A* with Opponent Awareness

- Finds shortest path to apple
- Adds +3 to +8 penalties near enemies
- Closed set prevents infinite loops
- O(n log n) complexity

2. 3-Step Prediction

Turn 1: Enemy at (10,10) moving right
Turn 2: Predicted (11,10)
Turn 3: Predicted (12,10) or (12,9) if aggressive
Our move: Position at (13,10) to intercept!

3. Territory Control

Scans 5-tile radius (25 tiles)
Counts tiles closer to us than enemies
+25 score per controlled tile

4. Escape Route Blocking

# = Wall, X = Us, E = Enemy (smaller)

# # # # #
# . E . #
# X # X #   Block left & bottom
# X X X #   Enemy trapped!
# # # # #

Bonus: +300 per blocked route

5. Dynamic Apple Priority

Value +5 to +1:  Aggressive pursuit (120x multiplier)
Value 0 to -3:   Eat if +3 length lead
Value  -4:      AVOID! (instant death)

Decision Flow

START  Emergency?  ESCAPE
       
   Apple Check  Poisonous?  AVOID
     
   Path Planning (BFS  Hamilton  A*  Spiral)
     
   Hunt Mode? (Size 5, weak enemy)  TRAP
     
   Score All Moves  SELECT BEST  OUTPUT

50ms Optimization

Before: 10ms decision, shallow 1-step analysis
After: 30ms decision, deep 3-step simulation

Extra time used for:

• 3-step prediction (was 1)
• Territory evaluation (new)
• Trap simulation (enhanced)
• Future space projection (new)

---

Official Game Mechanics

Snake Rules

• Movement: 1 square/turn (50ms), all simultaneous
• Death: Hit wall/body respawn after 1 turn
• Head-on: Both die, no kills
• Poison apple (-4): Instant death

Apple System

• Start value: +5 growth
• Decay: -0.1/turn (rounded up)
• Timeline:
  • Turns 0-49: +5 to +1 (GOOD)
  • Turns 50-79: 0 to -3 (RISKY)
  • Turn 80+: -4 or worse (DEADLY)

Scoring

1. Longest length achieved
2. Kill count (tiebreaker)
3. Snake index (final tiebreaker)

Configuration (snake_config.txt)

game_width      75      # Board width
game_height     75      # Board height
decay_rate      0.1     # Apple decay/turn
duration        300     # Game length (seconds)
speed           50      # Move interval (ms)
num_snakes      4       # Snakes per game

Game State Format

Init:

4 50 50 1
      Mode
    Height
  Width  
 Num snakes

Update:

8 16                                     Apple (8,16)
0                                        You are snake #0
alive 26 2 10,12 15,12 15,7 5,7 5,2     Snake 0
dead 6 6 14,13 19,13                    Snake 1
alive 2 1 12,13 12,14                   Snake 2
alive 17 1 31,14 21,14 15,14 15,13      Snake 3

Format: alive/dead length kills head,body... tail

Moves

0 = Up     | 4 = Turn Left (relative)
1 = Down   | 5 = Straight (relative)
2 = Left   | 6 = Turn Right (relative)
3 = Right  |

---

Setup Instructions

Prerequisites

1. Java JDK 21+ - Download
2. VS Code + Java Extension Pack
3. Git (optional)

Installation

# 1. Verify Java
java -version  # Should show 21+

# 2. Clone
git clone https://github.com/SindyMl/Snake-AI.git
cd Snake-AI

# 3. Open VS Code
code .

# 4. Build
.\build.ps1

# 5. Run
java -cp "src;lib/SnakeRunner.jar" MyAgent -develop

---

Screenshots

Local Testing

Tournament

---

Troubleshooting

"Unable to access jarfile"

cd Snake-AI
java -jar lib/SnakeRunner.jar ...  # Note: lib/ prefix

"Compilation failed"

java -version   # Check 21+
javac -version  # Should match

Security Manager Warnings

WARNING: The Security Manager is deprecated...

Normal! From game framework, not your code. Runs fine.

---

Resources

Documentation

• documentation/IMPLEMENTATION_SUMMARY.md
• documentation/ADVANCED_STRATEGIES.md
• documentation/50MS_OPTIMIZATIONS.md

Learning

• Java Tutorial
• A* Pathfinding

---

Acknowledgments

• Wits University - Snake AI Competition
• Community - Strategy discussions
• All opponents - Helping refine the AI

---

Made with for the Snake AI Tournament

Star this repo if helpful!