Evolution-ml

Running

App Files Files Community

Evolution-ml / index.html

openfree

Update index.html

85d6e29 verified 6 months ago

raw

history blame

60.3 kB

	// Feature to detect when we've completed the course
	function checkCourseCompletion() {
	const bestCar = getBestCar();

	if (bestCar && bestCar.checkpointIndex === track.checkpoints.length) {
	// Launch a bunch of celebratory confetti!
	createConfetti(100, canvas.width/2, canvas.height/2);
	setTimeout(() => createConfetti(50, canvas.width/4, canvas.height/2), 300);
	setTimeout(() => createConfetti(50, 3*canvas.width/4, canvas.height/2), 600);

	// Show victory message
	const message = document.createElement('div');
	message.style.position = 'absolute';
	message.style.top = '50%';
	message.style.left = '50%';
	message.style.transform = 'translate(-50%, -50%)';
	message.style.background = 'rgba(16, 185, 129, 0.9)';
	message.style.color = 'white';
	message.style.padding = '20px';
	message.style.borderRadius = '10px';
	message.style.fontSize = '24px';
	message.style.fontWeight = 'bold';
	message.style.boxShadow = '0 4px 6px rgba(0, 0, 0, 0.1)';
	message.style.zIndex = '1000';
	message.style.textAlign = 'center';
	message.innerHTML = `
	<div>🏆 Course Completed! 🏆</div>
	<div style="font-size: 16px; margin-top: 10px;">
	Generations: ${generation}<br>
	Fitness: ${Math.round(bestCar.fitness * 1000)}
	</div>
	<button id="continueBtn" style="
	background-color: white;
	color: #10b981;
	border: none;
	padding: 8px 16px;
	border-radius: 5px;
	margin-top: 15px;
	cursor: pointer;
	font-weight: bold;">
	Continue Training
	</button>
	`;

	document.body.appendChild(message);

	// Pause simulation
	isRunning = false;
	cancelAnimationFrame(animationId);

	// Event listener for the continue button
	document.getElementById('continueBtn').addEventListener('click', () => {
	document.body.removeChild(message);
	isRunning = true;
	lastUpdateTime = performance.now();
	animate();
	});
	}
	}<!DOCTYPE html>
	<html lang="en">
	<head>
	<meta charset="UTF-8">
	<meta name="viewport" content="width=device-width, initial-scale=1.0">
	<title>AI Driving Simulation</title>
	<link rel="stylesheet" href="https://cdnjs.cloudflare.com/ajax/libs/font-awesome/6.0.0/css/all.min.css" />
	<style>
	body {
	background-color: #111827;
	color: #f3f4f6;
	font-family: Arial, sans-serif;
	line-height: 1.5;
	margin: 0;
	padding: 20px;
	}

	.container {
	max-width: 1200px;
	margin: 0 auto;
	}

	h1 {
	color: #60a5fa;
	text-align: center;
	margin-bottom: 20px;
	font-family: 'Segoe UI', Tahoma, Geneva, Verdana, sans-serif;
	text-shadow: 0 2px 4px rgba(0, 0, 0, 0.3);
	letter-spacing: 1px;
	}

	canvas {
	background-color: #2d3748;
	border-radius: 12px;
	display: block;
	margin: 0 auto 20px;
	box-shadow: 0 10px 15px -3px rgba(0, 0, 0, 0.3), 0 4px 6px -2px rgba(0, 0, 0, 0.1);
	border: 2px solid #4a5568;
	}

	.controls {
	display: flex;
	gap: 10px;
	margin-bottom: 20px;
	justify-content: center;
	flex-wrap: wrap;
	}

	button {
	background-color: #3b82f6;
	color: white;
	border: none;
	padding: 10px 18px;
	border-radius: 8px;
	cursor: pointer;
	font-weight: bold;
	transition: all 0.2s ease;
	display: flex;
	align-items: center;
	gap: 5px;
	box-shadow: 0 4px 6px -1px rgba(0, 0, 0, 0.1);
	}

	button:hover {
	background-color: #2563eb;
	transform: translateY(-2px);
	box-shadow: 0 10px 15px -3px rgba(0, 0, 0, 0.1), 0 4px 6px -2px rgba(0, 0, 0, 0.05);
	}

	button:active {
	transform: translateY(1px);
	}

	.start-btn {
	background-color: #10b981;
	}

	.start-btn:hover {
	background-color: #059669;
	}

	.pause-btn {
	background-color: #f59e0b;
	}

	.pause-btn:hover {
	background-color: #d97706;
	}

	.reset-btn {
	background-color: #ef4444;
	}

	.reset-btn:hover {
	background-color: #dc2626;
	}

	.stats {
	display: flex;
	justify-content: space-between;
	background-color: #1f2937;
	padding: 18px;
	border-radius: 12px;
	margin-bottom: 20px;
	box-shadow: 0 4px 6px -1px rgba(0, 0, 0, 0.1);
	flex-wrap: wrap;
	gap: 10px;
	border: 1px solid #374151;
	}

	.stat-item {
	text-align: center;
	background-color: #2d3748;
	padding: 10px 15px;
	border-radius: 8px;
	min-width: 90px;
	transition: all 0.3s ease;
	}

	.stat-item:hover {
	transform: translateY(-2px);
	box-shadow: 0 4px 6px rgba(0, 0, 0, 0.1);
	}

	.stat-value {
	font-size: 1.4em;
	font-weight: bold;
	color: #60a5fa;
	margin-top: 5px;
	}

	.settings {
	background-color: #1f2937;
	padding: 20px;
	border-radius: 12px;
	margin-bottom: 20px;
	box-shadow: 0 4px 6px -1px rgba(0, 0, 0, 0.1);
	border: 1px solid #374151;
	}

	.slider-container {
	margin-bottom: 15px;
	background-color: #2d3748;
	padding: 12px;
	border-radius: 8px;
	}

	.slider-container label {
	display: block;
	margin-bottom: 8px;
	font-weight: 500;
	color: #d1d5db;
	}

	input[type="range"] {
	width: 100%;
	margin-bottom: 8px;
	height: 6px;
	-webkit-appearance: none;
	background: #4b5563;
	border-radius: 5px;
	outline: none;
	}

	input[type="range"]::-webkit-slider-thumb {
	-webkit-appearance: none;
	width: 18px;
	height: 18px;
	background: #3b82f6;
	border-radius: 50%;
	cursor: pointer;
	transition: background 0.2s;
	}

	input[type="range"]::-webkit-slider-thumb:hover {
	background: #2563eb;
	}

	.progress-container {
	background-color: #374151;
	height: 10px;
	border-radius: 5px;
	margin-top: 10px;
	overflow: hidden;
	box-shadow: inset 0 1px 3px rgba(0, 0, 0, 0.2);
	}

	.progress-bar {
	height: 100%;
	background: linear-gradient(90deg, #10b981, #3b82f6);
	width: 0;
	transition: width 0.3s;
	box-shadow: 0 0 5px rgba(16, 185, 129, 0.5);
	}

	/* Car emoji styling */
	.car-emoji {
	font-size: 1.2em;
	margin-right: 5px;
	}

	/* Added visual elements */
	.section-title {
	display: flex;
	align-items: center;
	margin-bottom: 15px;
	border-bottom: 1px solid #374151;
	padding-bottom: 8px;
	}

	.section-title i {
	margin-right: 8px;
	color: #60a5fa;
	}

	/* Custom checkbox */
	.toggle-switch {
	position: relative;
	display: inline-block;
	width: 50px;
	height: 24px;
	}

	.toggle-switch input {
	opacity: 0;
	width: 0;
	height: 0;
	}

	.toggle-slider {
	position: absolute;
	cursor: pointer;
	top: 0;
	left: 0;
	right: 0;
	bottom: 0;
	background-color: #4b5563;
	transition: .4s;
	border-radius: 24px;
	}

	.toggle-slider:before {
	position: absolute;
	content: "";
	height: 16px;
	width: 16px;
	left: 4px;
	bottom: 4px;
	background-color: white;
	transition: .4s;
	border-radius: 50%;
	}

	input:checked + .toggle-slider {
	background-color: #3b82f6;
	}

	input:checked + .toggle-slider:before {
	transform: translateX(26px);
	}
	</style>
	</head>
	<body>
	<div class="container">
	<h1>🏎️ AI Driving Simulation</h1>

	<canvas id="simulationCanvas" width="800" height="500"></canvas>

	<div class="controls">
	<button id="startBtn" class="start-btn"><i class="fas fa-play"></i> Start</button>
	<button id="pauseBtn" class="pause-btn"><i class="fas fa-pause"></i> Pause</button>
	<button id="resetBtn" class="reset-btn"><i class="fas fa-sync-alt"></i> New Track</button>
	<button id="saveBtn"><i class="fas fa-save"></i> Save Model</button>
	<button id="loadBtn"><i class="fas fa-upload"></i> Load Model</button>
	</div>

	<div class="stats">
	<div class="stat-item">
	<div><i class="fas fa-dna"></i> Generation</div>
	<div id="generationCount" class="stat-value">0</div>
	</div>
	<div class="stat-item">
	<div><i class="fas fa-car"></i> Alive</div>
	<div class="stat-value"><span id="aliveCount">0</span>/<span id="populationCount">0</span></div>
	</div>
	<div class="stat-item">
	<div><i class="fas fa-trophy"></i> Best Fitness</div>
	<div id="maxFitness" class="stat-value">0</div>
	</div>
	<div class="stat-item">
	<div><i class="fas fa-tachometer-alt"></i> FPS</div>
	<div id="fpsCounter" class="stat-value">0</div>
	</div>
	</div>

	<div class="settings">
	<div class="section-title">
	<i class="fas fa-sliders-h"></i> <h3>Simulation Settings</h3>
	</div>

	<div class="slider-container">
	<label for="populationSlider"><i class="fas fa-users"></i> Population Size:</label>
	<input type="range" id="populationSlider" min="10" max="300" value="100">
	<span id="populationValue">100</span>
	</div>

	<div class="slider-container">
	<label for="mutationSlider"><i class="fas fa-random"></i> Mutation Rate:</label>
	<input type="range" id="mutationSlider" min="1" max="100" value="10">
	<span id="mutationValue">10%</span>
	</div>

	<div class="slider-container">
	<label for="speedSlider"><i class="fas fa-fast-forward"></i> Simulation Speed:</label>
	<input type="range" id="speedSlider" min="1" max="30" value="5">
	<span id="speedValue">15x</span>
	</div>

	<div class="slider-container">
	<label><i class="fas fa-flag-checkered"></i> Track Progress:</label>
	<div class="progress-container">
	<div id="bestProgressBar" class="progress-bar"></div>
	</div>
	</div>
	</div>

	<div class="settings">
	<div class="section-title">
	<i class="fas fa-info-circle"></i> <h3>About This Simulation</h3>
	</div>
	<p>This simulation demonstrates how AI can learn to drive using genetic algorithms and neural networks. Cars must navigate randomly generated tracks without any prior knowledge of the environment.</p>
	<p><strong>Key Improvements:</strong></p>
	<ul>
	<li><i class="fas fa-brain"></i> <strong>Enhanced Neural Network:</strong> Using Sigmoid activation function for smoother decision making</li>
	<li><i class="fas fa-random"></i> <strong>Crossover:</strong> Combining the best traits from parent models</li>
	<li><i class="fas fa-chart-line"></i> <strong>Adaptive Mutation:</strong> Automatically adjusts as generations progress</li>
	<li><i class="fas fa-bolt"></i> <strong>Performance Optimization:</strong> Delta-time based updates for consistent simulation</li>
	<li><i class="fas fa-exclamation-triangle"></i> <strong>Improved Collision Detection:</strong> More accurate polygon-based detection</li>
	<li><i class="fas fa-save"></i> <strong>Model Saving:</strong> Save and load your best models</li>
	</ul>
	</div>
	</div>

	<script>
	document.addEventListener('DOMContentLoaded', () => {
	// Canvas 설정
	const canvas = document.getElementById('simulationCanvas');
	const ctx = canvas.getContext('2d');

	// UI 요소
	const startBtn = document.getElementById('startBtn');
	const pauseBtn = document.getElementById('pauseBtn');
	const resetBtn = document.getElementById('resetBtn');
	const saveBtn = document.getElementById('saveBtn');
	const loadBtn = document.getElementById('loadBtn');

	const populationSlider = document.getElementById('populationSlider');
	const mutationSlider = document.getElementById('mutationSlider');
	const speedSlider = document.getElementById('speedSlider');

	const populationValue = document.getElementById('populationValue');
	const mutationValue = document.getElementById('mutationValue');
	const speedValue = document.getElementById('speedValue');

	const generationCount = document.getElementById('generationCount');
	const aliveCount = document.getElementById('aliveCount');
	const populationCount = document.getElementById('populationCount');
	const maxFitness = document.getElementById('maxFitness');
	const fpsCounter = document.getElementById('fpsCounter');
	const bestProgressBar = document.getElementById('bestProgressBar');

	// 시뮬레이션 매개변수
	let populationSize = parseInt(populationSlider.value);
	let mutationRate = parseInt(mutationSlider.value) / 100;
	let simulationSpeed = parseInt(speedSlider.value) * 3; // 3배 빠른 속도
	let isRunning = false;
	let generation = 0;
	let fps = 0;
	let bestCarProgress = 0;
	let deltaTime = 0;
	let lastUpdateTime = 0;
	let frameCount = 0;
	let lastFpsUpdate = 0;

	// 활성화 함수
	const sigmoid = (x) => 1 / (1 + Math.exp(-x));
	const relu = (x) => Math.max(0, x);

	// 트랙 정의
	const track = {
	walls: [],
	checkpoints: [],
	startPosition: { x: 100, y: 250, angle: 0 },

	generateRandomTrack() {
	this.walls = [];
	this.checkpoints = [];

	// 외부 경계 벽 (항상 존재)
	this.walls.push(
	{ x: 50, y: 50, width: 700, height: 20 }, // 상단
	{ x: 50, y: 50, width: 20, height: 400 }, // 좌측
	{ x: 50, y: 430, width: 700, height: 20 }, // 하단
	{ x: 730, y: 50, width: 20, height: 400 } // 우측
	);

	// 무작위 장애물 생성
	const obstacleCount = 3 + Math.floor(Math.random() * 6);
	for (let i = 0; i < obstacleCount; i++) {
	const isVertical = Math.random() > 0.5;
	let x, y, width, height;

	if (isVertical) {
	width = 20;
	height = 50 + Math.random() * 200;
	x = 100 + Math.random() * 600;
	y = 100 + Math.random() * (400 - height);
	} else {
	width = 50 + Math.random() * 200;
	height = 20;
	x = 100 + Math.random() * (700 - width);
	y = 100 + Math.random() * 300;
	}

	// 시작 위치를 막지 않도록 확인
	if (!(x < 150 && y < 300 && y + height > 200)) {
	this.walls.push({ x, y, width, height });
	}
	}

	// 체크포인트 생성
	const checkpointCount = 3 + Math.floor(Math.random() * 3);
	const checkpointSize = 30;

	// 장애물 주변을 통과해야 하는 위치 생성
	const possiblePositions = [
	{ x: 700, y: 100 }, // 우측 상단
	{ x: 600, y: 400 }, // 우측 하단 중앙
	{ x: 300, y: 400 }, // 하단 중앙
	{ x: 100, y: 300 }, // 좌측 중앙
	{ x: 400, y: 100 }, // 상단 중앙
	{ x: 200, y: 200 }, // 좌측 중앙
	{ x: 600, y: 200 } // 우측 중앙
	];

	// 셔플하고 체크포인트 수만큼 선택
	const shuffled = [...possiblePositions].sort(() => 0.5 - Math.random());
	for (let i = 0; i < checkpointCount; i++) {
	const pos = shuffled[i];
	this.checkpoints.push({
	x: pos.x,
	y: pos.y,
	width: checkpointSize,
	height: checkpointSize
	});
	}

	// 시작 위치 설정 (항상 좌측, 수직 위치는 무작위)
	this.startPosition = {
	x: 100,
	y: 100 + Math.random() * 300,
	angle: 0
	};
	},

	draw(ctx) {
	// 벽 그리기
	ctx.fillStyle = '#4a5568';
	this.walls.forEach(wall => {
	ctx.fillRect(wall.x, wall.y, wall.width, wall.height);
	});

	// 체크포인트 그리기
	ctx.fillStyle = 'rgba(74, 222, 128, 0.3)';
	this.checkpoints.forEach((checkpoint, index) => {
	ctx.fillRect(checkpoint.x, checkpoint.y, checkpoint.width, checkpoint.height);

	// Add checkpoint number
	ctx.fillStyle = 'white';
	ctx.font = '12px Arial';
	ctx.textAlign = 'center';
	ctx.textBaseline = 'middle';
	ctx.fillText((index + 1).toString(),
	checkpoint.x + checkpoint.width/2,
	checkpoint.y + checkpoint.height/2);

	ctx.fillStyle = 'rgba(74, 222, 128, 0.3)';
	});

	// 시작 위치 그리기
	ctx.fillStyle = 'rgba(96, 165, 250, 0.5)';
	ctx.fillRect(this.startPosition.x - 15, this.startPosition.y - 25, 30, 50);

	// Draw start flag
	ctx.fillStyle = 'white';
	ctx.font = '14px Arial';
	ctx.textAlign = 'center';
	ctx.fillText("START", this.startPosition.x, this.startPosition.y + 15);
	}
	};

	// 자동차 클래스
	class Car {
	constructor(brain) {
	this.reset();
	this.brain = brain ? brain : new NeuralNetwork([5, 8, 2]);
	this.fitness = 0;
	this.checkpointIndex = 0;
	this.sensors = [0, 0, 0, 0, 0]; // 전방, 좌측, 우측, 좌전방, 우전방
	this.sensorAngles = [0, -Math.PI/4, Math.PI/4, -Math.PI/8, Math.PI/8];
	this.sensorLength = 100;
	this.color = 'rgba(59, 130, 246, 0.8)';
	this.isBest = false;
	this.lastPosition = { x: 0, y: 0 };
	this.stuckTime = 0; // 자동차가 움직이지 않는 시간 추적
	}

	reset() {
	this.x = track.startPosition.x;
	this.y = track.startPosition.y;
	this.angle = track.startPosition.angle;
	this.speed = 0;
	this.maxSpeed = 10; // 최대 속도 증가
	this.acceleration = 0.2; // 가속도 증가
	this.rotationSpeed = 0.1; // 회전 속도 증가
	this.damaged = false;
	this.checkpointIndex = 0;
	this.fitness = 0;
	this.stuckTime = 0;
	this.lastPosition = { x: this.x, y: this.y };
	}

	update(dt) {
	if (this.damaged) return;

	// 이전 위치 저장
	this.lastPosition = { x: this.x, y: this.y };

	// 센서 업데이트
	this.updateSensors();

	// 신경망 출력 가져오기
	const outputs = this.brain.predict(this.sensors);

	// 조향 적용 (outputs[0] = 왼쪽, outputs[1] = 오른쪽)
	const steering = outputs[1] - outputs[0]; // -1에서 1 사이
	this.angle += steering * this.rotationSpeed * dt;

	// 속도와 위치 업데이트
	this.speed = this.maxSpeed;
	this.x += Math.sin(this.angle) * this.speed * dt;
	this.y -= Math.cos(this.angle) * this.speed * dt;

	// 충돌 검사
	this.checkCollisions();

	// 체크포인트 검사
	this.checkCheckpoints();

	// 정지 확인 (자동차가 움직이지 않는 경우)
	const distance = Math.sqrt(
	Math.pow(this.x - this.lastPosition.x, 2) +
	Math.pow(this.y - this.lastPosition.y, 2)
	);

	if (distance < 0.5 * dt) {
	this.stuckTime += dt;
	if (this.stuckTime > 1.5) { // 정지 판단 시간 단축 (3초 → 1.5초)
	this.damaged = true;
	}
	} else {
	this.stuckTime = 0;

	// 적합도 업데이트
	this.fitness += distance; // 이동 거리에 따른 적합도
	}
	}

	updateSensors() {
	this.sensors = this.sensorAngles.map(angle => {
	const sensorAngle = this.angle + angle;
	let sensorEndX = this.x + Math.sin(sensorAngle) * this.sensorLength;
	let sensorEndY = this.y - Math.cos(sensorAngle) * this.sensorLength;

	let minDistance = this.sensorLength;

	// 모든 벽에 대해 검사
	for (const wall of track.walls) {
	const intersection = this.lineRectIntersection(
	this.x, this.y, sensorEndX, sensorEndY,
	wall.x, wall.y, wall.width, wall.height
	);

	if (intersection) {
	const distance = Math.sqrt(
	Math.pow(intersection.x - this.x, 2) +
	Math.pow(intersection.y - this.y, 2)
	);

	minDistance = Math.min(minDistance, distance);
	}
	}

	// 정규화된 거리 반환 (0-1 범위; 1 = 장애물 없음, 0 = 자동차 바로 앞에 장애물)
	return 1 - (minDistance / this.sensorLength);
	});
	}

	lineRectIntersection(x1, y1, x2, y2, rx, ry, rw, rh) {
	// 직선이 직사각형의 변과 교차하는지 확인
	const left = this.lineLineIntersection(x1, y1, x2, y2, rx, ry, rx, ry + rh);
	const right = this.lineLineIntersection(x1, y1, x2, y2, rx + rw, ry, rx + rw, ry + rh);
	const top = this.lineLineIntersection(x1, y1, x2, y2, rx, ry, rx + rw, ry);
	const bottom = this.lineLineIntersection(x1, y1, x2, y2, rx, ry + rh, rx + rw, ry + rh);

	let closestIntersection = null;
	let minDistance = Infinity;

	[left, right, top, bottom].forEach(intersection => {
	if (intersection) {
	const distance = Math.sqrt(Math.pow(intersection.x - x1, 2) + Math.pow(intersection.y - y1, 2));
	if (distance < minDistance) {
	minDistance = distance;
	closestIntersection = intersection;
	}
	}
	});

	return closestIntersection;
	}

	lineLineIntersection(x1, y1, x2, y2, x3, y3, x4, y4) {
	// 두 직선 사이의 교차점 계산
	const denominator = (y4 - y3) * (x2 - x1) - (x4 - x3) * (y2 - y1);

	if (denominator === 0) return null; // 직선이 평행

	const ua = ((x4 - x3) * (y1 - y3) - (y4 - y3) * (x1 - x3)) / denominator;
	const ub = ((x2 - x1) * (y1 - y3) - (y2 - y1) * (x1 - x3)) / denominator;

	if (ua >= 0 && ua <= 1 && ub >= 0 && ub <= 1) {
	return {
	x: x1 + ua * (x2 - x1),
	y: y1 + ua * (y2 - y1)
	};
	}

	return null;
	}

	checkCollisions() {
	// 벽과의 충돌 감지 (개선된 계산)
	// 자동차를 다각형으로 표현하여 더 정확한 충돌 감지
	const carCorners = this.getCarCorners();

	// 모든 벽에 대해 검사
	for (const wall of track.walls) {
	// 간단한 사각형 충돌 확인 (최적화를 위한 첫 단계)
	if (this.x > wall.x - 10 && this.x < wall.x + wall.width + 10 &&
	this.y > wall.y - 10 && this.y < wall.y + wall.height + 10) {

	// 자동차 꼭지점이 벽 내부에 있는지 확인
	for (const corner of carCorners) {
	if (corner.x > wall.x && corner.x < wall.x + wall.width &&
	corner.y > wall.y && corner.y < wall.y + wall.height) {
	this.damaged = true;
	return;
	}
	}
	}
	}

	// 경계 확인
	if (this.x < 0 \|\| this.x > canvas.width \|\| this.y < 0 \|\| this.y > canvas.height) {
	this.damaged = true;
	}
	}

	getCarCorners() {
	// 자동차의 네 꼭지점 계산
	const width = 12;
	const height = 20;
	const cornerOffsets = [
	{ x: -width/2, y: -height/2 }, // 좌상단
	{ x: width/2, y: -height/2 }, // 우상단
	{ x: width/2, y: height/2 }, // 우하단
	{ x: -width/2, y: height/2 } // 좌하단
	];

	return cornerOffsets.map(offset => {
	const rotatedX = offset.x * Math.cos(this.angle) - offset.y * Math.sin(this.angle);
	const rotatedY = offset.x * Math.sin(this.angle) + offset.y * Math.cos(this.angle);
	return {
	x: this.x + rotatedX,
	y: this.y + rotatedY
	};
	});
	}

	checkCheckpoints() {
	if (this.checkpointIndex >= track.checkpoints.length) return;

	const checkpoint = track.checkpoints[this.checkpointIndex];
	if (this.x > checkpoint.x && this.x < checkpoint.x + checkpoint.width &&
	this.y > checkpoint.y && this.y < checkpoint.y + checkpoint.height) {
	this.checkpointIndex++;
	this.fitness += 1000; // Bonus for reaching checkpoint

	// Update best progress visualization
	const progress = this.checkpointIndex / track.checkpoints.length;
	if (progress > bestCarProgress) {
	bestCarProgress = progress;
	bestProgressBar.style.width = `${progress * 100}%`;

	// Add confetti effect for completed checkpoints
	if (this.checkpointIndex > 0) {
	createConfetti(10, this.x, this.y);
	}
	}
	}
	}

	draw(ctx) {
	if (this.damaged) return;

	ctx.save();
	ctx.translate(this.x, this.y);
	ctx.rotate(this.angle);

	// Draw car body - Enhanced car shape with emoji style
	if (this.isBest) {
	// Draw fancy car for the best performer
	ctx.fillStyle = 'rgba(220, 38, 38, 0.9)';

	// Main body
	ctx.beginPath();
	ctx.roundRect(-6, -10, 12, 20, 2);
	ctx.fill();

	// Wheels
	ctx.fillStyle = '#000';
	ctx.fillRect(-7, -8, 2, 4); // left front
	ctx.fillRect(5, -8, 2, 4); // right front
	ctx.fillRect(-7, 4, 2, 4); // left rear
	ctx.fillRect(5, 4, 2, 4); // right rear

	// Windshield
	ctx.fillStyle = '#60a5fa';
	ctx.beginPath();
	ctx.roundRect(-4, -8, 8, 6, 1);
	ctx.fill();

	// Draw a small crown on top
	ctx.fillStyle = '#facc15';
	ctx.beginPath();
	ctx.moveTo(-3, -11);
	ctx.lineTo(-1, -13);
	ctx.lineTo(1, -11);
	ctx.lineTo(3, -13);
	ctx.lineTo(3, -10);
	ctx.lineTo(-3, -10);
	ctx.fill();
	} else {
	// Regular car
	ctx.fillStyle = this.color;

	// Main body
	ctx.beginPath();
	ctx.roundRect(-6, -10, 12, 20, 2);
	ctx.fill();

	// Wheels (simple)
	ctx.fillStyle = '#000';
	ctx.fillRect(-7, -7, 2, 3); // left front
	ctx.fillRect(5, -7, 2, 3); // right front
	ctx.fillRect(-7, 4, 2, 3); // left rear
	ctx.fillRect(5, 4, 2, 3); // right rear

	// Simple windshield
	ctx.fillStyle = '#a3e0ff';
	ctx.fillRect(-4, -7, 8, 5);
	}

	// 최고 자동차의 센서 그리기
	if (this.isBest) {
	ctx.restore(); // 컨텍스트 복원

	// 센서 레이 그리기
	ctx.strokeStyle = 'rgba(255, 255, 255, 0.5)';
	ctx.lineWidth = 1;

	this.sensorAngles.forEach((angle, i) => {
	const sensorAngle = this.angle + angle;
	const sensorValue = this.sensors[i];
	const sensorLength = this.sensorLength * (1 - sensorValue);

	const endX = this.x + Math.sin(sensorAngle) * sensorLength;
	const endY = this.y - Math.cos(sensorAngle) * sensorLength;

	ctx.beginPath();
	ctx.moveTo(this.x, this.y);
	ctx.lineTo(endX, endY);
	ctx.stroke();
	});

	return; // 이미 ctx.restore()를 호출했으므로 여기서 종료
	}

	ctx.restore();
	}

	clone() {
	return new Car(this.brain.clone());
	}
	}

	// 신경망 클래스
	class NeuralNetwork {
	constructor(neuronCounts) {
	this.levels = [];
	for (let i = 0; i < neuronCounts.length - 1; i++) {
	this.levels.push(new Level(
	neuronCounts[i], neuronCounts[i + 1]
	));
	}
	}

	predict(givenInputs) {
	let outputs = Level.feedForward(
	givenInputs, this.levels[0]
	);

	for (let i = 1; i < this.levels.length; i++) {
	outputs = Level.feedForward(
	outputs, this.levels[i]
	);
	}

	return outputs;
	}

	clone() {
	const clone = new NeuralNetwork([]);
	clone.levels = this.levels.map(level => level.clone());
	return clone;
	}

	mutate(rate) {
	for (const level of this.levels) {
	for (let i = 0; i < level.biases.length; i++) {
	if (Math.random() < rate) {
	level.biases[i] = lerp(
	level.biases[i],
	Math.random() * 2 - 1,
	0.5
	);
	}
	}
	for (let i = 0; i < level.weights.length; i++) {
	for (let j = 0; j < level.weights[i].length; j++) {
	if (Math.random() < rate) {
	level.weights[i][j] = lerp(
	level.weights[i][j],
	Math.random() * 2 - 1,
	0.5
	);
	}
	}
	}
	}
	}

	static crossover(parentA, parentB) {
	// 두 부모 신경망에서 새 신경망 생성
	if (parentA.levels.length !== parentB.levels.length) {
	console.error("부모 신경망 구조가 다릅니다!");
	return parentA.clone();
	}

	const childNetwork = new NeuralNetwork([]);
	childNetwork.levels = [];

	for (let l = 0; l < parentA.levels.length; l++) {
	const levelA = parentA.levels[l];
	const levelB = parentB.levels[l];

	if (levelA.inputs.length !== levelB.inputs.length \|\|
	levelA.outputs.length !== levelB.outputs.length) {
	console.error("부모 레벨 구조가 다릅니다!");
	return parentA.clone();
	}

	const childLevel = new Level(levelA.inputs.length, levelA.outputs.length);

	// 교차점 선택 (단일점 교차)
	const biasesSwitch = Math.floor(Math.random() * levelA.biases.length);

	// 바이어스 교차
	for (let i = 0; i < childLevel.biases.length; i++) {
	childLevel.biases[i] = i < biasesSwitch
	? levelA.biases[i]
	: levelB.biases[i];
	}

	// 가중치 교차
	for (let i = 0; i < childLevel.weights.length; i++) {
	const weightSwitch = Math.floor(Math.random() * levelA.weights[i].length);

	for (let j = 0; j < childLevel.weights[i].length; j++) {
	childLevel.weights[i][j] = j < weightSwitch
	? levelA.weights[i][j]
	: levelB.weights[i][j];
	}
	}

	childNetwork.levels.push(childLevel);
	}

	return childNetwork;
	}

	toJSON() {
	return {
	levels: this.levels.map(level => ({
	inputs: level.inputs,
	outputs: level.outputs,
	biases: level.biases,
	weights: level.weights
	}))
	};
	}

	static fromJSON(data) {
	const network = new NeuralNetwork([]);
	network.levels = data.levels.map(levelData => {
	const level = new Level(levelData.inputs.length, levelData.outputs.length);
	level.inputs = [...levelData.inputs];
	level.outputs = [...levelData.outputs];
	level.biases = [...levelData.biases];
	level.weights = levelData.weights.map(w => [...w]);
	return level;
	});
	return network;
	}
	}

	function lerp(a, b, t) {
	return a + (b - a) * t;
	}

	class Level {
	constructor(inputCount, outputCount) {
	this.inputs = new Array(inputCount);
	this.outputs = new Array(outputCount);
	this.biases = new Array(outputCount);
	this.weights = [];

	for (let i = 0; i < inputCount; i++) {
	this.weights[i] = new Array(outputCount);
	}

	Level.#randomize(this);
	}

	static #randomize(level) {
	for (let i = 0; i < level.inputs.length; i++) {
	for (let j = 0; j < level.outputs.length; j++) {
	level.weights[i][j] = Math.random() * 2 - 1;
	}
	}

	for (let i = 0; i < level.biases.length; i++) {
	level.biases[i] = Math.random() * 2 - 1;
	}
	}

	static feedForward(givenInputs, level) {
	// 입력 값 설정
	for (let i = 0; i < level.inputs.length; i++) {
	level.inputs[i] = givenInputs[i];
	}

	// 각 출력 뉴런에 대해 가중 합계 계산
	for (let i = 0; i < level.outputs.length; i++) {
	let sum = 0;
	for (let j = 0; j < level.inputs.length; j++) {
	sum += level.inputs[j] * level.weights[j][i];
	}

	// Sigmoid 활성화 함수 적용
	level.outputs[i] = sigmoid(sum - level.biases[i]);
	}

	return level.outputs;
	}

	clone() {
	const clone = new Level(this.inputs.length, this.outputs.length);
	clone.inputs = [...this.inputs];
	clone.outputs = [...this.outputs];
	clone.biases = [...this.biases];
	clone.weights = this.weights.map(arr => [...arr]);
	return clone;
	}
	}

	// 유전 알고리즘 함수
	function nextGeneration() {
	generation++;
	generationCount.textContent = generation;

	// 적합도 계산
	calculateFitness();

	// 새 인구 생성
	const newPopulation = [];

	// 적응형 돌연변이율 적용
	const progressRate = bestCarProgress;
	const adaptedRate = mutationRate * (1 - progressRate * 0.5);
	mutationRate = Math.max(0.01, adaptedRate); // 최소 1%
	mutationValue.textContent = `${Math.round(mutationRate * 100)}%`;
	mutationSlider.value = Math.round(mutationRate * 100);

	// 이전 세대에서 최고 자동차 추가 (엘리티즘)
	const eliteCount = Math.max(1, Math.floor(populationSize * 0.05)); // 5% 엘리트
	const eliteCars = getTopCars(eliteCount);

	for (const eliteCar of eliteCars) {
	eliteCar.isBest = eliteCar === eliteCars[0];
	newPopulation.push(eliteCar.clone());
	}

	// 교차와 돌연변이로 나머지 채우기
	while (newPopulation.length < populationSize) {
	if (Math.random() < 0.7 && newPopulation.length + 1 < populationSize) {
	// 교차
	const parentA = selectParent();
	const parentB = selectParent();
	const child = new Car(NeuralNetwork.crossover(parentA.brain, parentB.brain));

	// 자식에게 약간의 돌연변이 적용
	child.brain.mutate(mutationRate);
	newPopulation.push(child);
	} else {
	// 돌연변이만
	const parent = selectParent();
	const child = parent.clone();
	child.brain.mutate(mutationRate);
	newPopulation.push(child);
	}
	}

	// 이전 인구 교체
	cars = newPopulation;

	// 자동차 초기화
	cars.forEach(car => car.reset());

	// 진행률 초기화
	bestCarProgress = 0;
	bestProgressBar.style.width = '0%';
	}

	function calculateFitness() {
	let sum = 0;
	let max = 0;

	cars.forEach(car => {
	// 체크포인트 달성 보너스
	car.fitness += car.checkpointIndex * 500;

	sum += car.fitness;
	if (car.fitness > max) max = car.fitness;
	});

	// 적합도 정규화
	cars.forEach(car => {
	car.fitness = car.fitness / sum;
	});

	// UI 업데이트
	maxFitness.textContent = Math.round(max);
	}

	function getTopCars(count) {
	return [...cars]
	.sort((a, b) => b.fitness - a.fitness)
	.slice(0, count);
	}

	function getBestCar() {
	let bestCar = cars[0];
	let bestFitness = cars[0].fitness;

	for (let i = 1; i < cars.length; i++) {
	if (cars[i].fitness > bestFitness) {
	bestFitness = cars[i].fitness;
	bestCar = cars[i];
	}
	}

	return bestCar;
	}

	function selectParent() {
	// 룰렛 휠 선택
	let index = 0;
	let r = Math.random();

	while (r > 0 && index < cars.length) {
	r -= cars[index].fitness;
	index++;
	}

	index = Math.min(cars.length - 1, Math.max(0, index - 1));
	return cars[index];
	}

	// 모델 저장/불러오기 함수
	function saveBestModel() {
	const bestCar = getBestCar();
	if (bestCar) {
	try {
	const modelData = {
	brain: bestCar.brain.toJSON(),
	fitness: bestCar.fitness,
	generation: generation,
	timestamp: new Date().toISOString()
	};

	localStorage.setItem('bestCarModel', JSON.stringify(modelData));
	return true;
	} catch (error) {
	console.error('Error saving model:', error);
	return false;
	}
	}
	return false;
	}

	function loadModel() {
	try {
	const savedModel = localStorage.getItem('bestCarModel');
	if (savedModel) {
	const modelData = JSON.parse(savedModel);

	// Use saved model for new population
	const newPopulation = [];

	// Create best car with restored brain
	const restoredBrain = NeuralNetwork.fromJSON(modelData.brain);
	const bestCar = new Car(restoredBrain);
	bestCar.isBest = true;
	newPopulation.push(bestCar);

	// Create variants from this model to fill population
	for (let i = 1; i < populationSize; i++) {
	const car = bestCar.clone();
	car.brain.mutate(mutationRate);
	newPopulation.push(car);
	}

	// Replace population
	cars = newPopulation;

	// Reset cars
	cars.forEach(car => car.reset());

	// Create a celebratory confetti effect
	createConfetti(50, canvas.width/2, canvas.height/2);

	return true;
	}
	} catch (error) {
	console.error('Error loading model:', error);
	}
	return false;
	}

	// 시뮬레이션 상태
	let cars = [];
	let animationId;

	// 시뮬레이션 초기화
	function init() {
	// 무작위 트랙 생성
	track.generateRandomTrack();

	// 초기 인구 생성
	cars = [];
	for (let i = 0; i < populationSize; i++) {
	cars.push(new Car());
	}

	// 통계 초기화
	generation = 0;
	generationCount.textContent = generation;
	populationCount.textContent = populationSize;

	// 시뮬레이션 시작
	isRunning = true;
	lastUpdateTime = performance.now();
	init();
	}

	// 결과 업데이트 속도 제한 (매 프레임마다 하지 않고 10프레임마다 한 번씩)
	let updateFrameCount = 0;

	// Confetti particle system
	const confetti = [];

	function createConfetti(count, x, y) {
	for (let i = 0; i < count; i++) {
	confetti.push({
	x: x,
	y: y,
	size: 3 + Math.random() * 5,
	color: `hsl(${Math.random() * 360}, 100%, 70%)`,
	vx: -2 + Math.random() * 4,
	vy: -3 - Math.random() * 2,
	gravity: 0.1,
	life: 1, // 1 = full life, 0 = dead
	maxLife: 1 + Math.random()
	});
	}
	}

	function updateConfetti(dt) {
	for (let i = confetti.length - 1; i >= 0; i--) {
	const particle = confetti[i];

	// Update position
	particle.x += particle.vx * dt * 60;
	particle.y += particle.vy * dt * 60;
	particle.vy += particle.gravity * dt * 60;

	// Update life
	particle.life -= 0.016 * dt * 60;

	// Remove dead particles
	if (particle.life <= 0) {
	confetti.splice(i, 1);
	}
	}
	}

	function drawConfetti(ctx) {
	for (const particle of confetti) {
	ctx.fillStyle = particle.color;
	ctx.globalAlpha = particle.life;
	ctx.fillRect(
	particle.x - particle.size/2,
	particle.y - particle.size/2,
	particle.size,
	particle.size
	);
	}
	ctx.globalAlpha = 1;
	}

	// 주요 애니메이션 루프
	function animate(currentTime = 0) {
	if (!isRunning) return;

	animationId = requestAnimationFrame(animate);

	// 델타 타임 계산
	deltaTime = (currentTime - lastUpdateTime) / 1000; // 초 단위
	lastUpdateTime = currentTime;

	// 시뮬레이션 속도 적용
	deltaTime *= simulationSpeed;

	// 최대 델타 타임 제한 (시뮬레이션 안정성을 위해)
	deltaTime = Math.min(deltaTime, 0.2);

	// FPS 계산 (10프레임마다 업데이트)
	frameCount++;
	updateFrameCount++;
	if (currentTime - lastFpsUpdate >= 1000) {
	fps = Math.round((frameCount * 1000) / (currentTime - lastFpsUpdate));
	if (updateFrameCount >= 10) {
	fpsCounter.textContent = fps;
	updateFrameCount = 0;
	}
	frameCount = 0;
	lastFpsUpdate = currentTime;
	}

	// 캔버스 지우기
	ctx.clearRect(0, 0, canvas.width, canvas.height);

	// 트랙 그리기
	track.draw(ctx);

	// 자동차 업데이트 및 그리기
	let alive = 0;
	cars.forEach(car => {
	car.update(deltaTime);
	car.draw(ctx);
	if (!car.damaged) alive++;
	});

	// Draw confetti particles
	updateConfetti(deltaTime);
	drawConfetti(ctx);

	aliveCount.textContent = alive;

	// 모든 자동차가 손상되었는지 확인
	if (alive === 0) {
	nextGeneration();
	}

	// 최고 자동차 강조 표시
	const bestCar = getBestCar();
	if (bestCar) {
	bestCar.isBest = true;
	bestCar.color = 'rgba(220, 38, 38, 0.9)';
	}
	}

	// 이벤트 리스너
	startBtn.addEventListener('click', () => {
	if (!isRunning) {
	isRunning = true;
	lastUpdateTime = performance.now();
	animate();
	}
	});

	pauseBtn.addEventListener('click', () => {
	isRunning = false;
	cancelAnimationFrame(animationId);
	});

	resetBtn.addEventListener('click', () => {
	isRunning = false;
	cancelAnimationFrame(animationId);
	init();
	});

	saveBtn.addEventListener('click', () => {
	if (saveBestModel()) {
	alert('Model saved successfully!');
	} else {
	alert('Error saving model');
	}
	});

	loadBtn.addEventListener('click', () => {
	if (loadModel()) {
	alert('Model loaded successfully!');
	} else {
	alert('No saved model found or error loading model');
	}
	});

	// 슬라이더 이벤트 리스너
	populationSlider.addEventListener('input', () => {
	populationSize = parseInt(populationSlider.value);
	populationValue.textContent = populationSize;
	populationCount.textContent = populationSize;
	});

	mutationSlider.addEventListener('input', () => {
	mutationRate = parseInt(mutationSlider.value) / 100;
	mutationValue.textContent = `${parseInt(mutationSlider.value)}%`;
	});

	speedSlider.addEventListener('input', () => {
	simulationSpeed = parseInt(speedSlider.value) * 3;
	speedValue.textContent = `${simulationSpeed}x`;
	});

	// 시뮬레이션 초기화 및 시작
	init();
	});
	</script>
	</body>
	</html>