--- id: [[P-Reinforce|P-Reinforce]]-AI-PATHPLAN category: "10_Wiki/πŸ’‘ Topics/AI" confidence_score: 0.98 tags: [Path Planning, A* Algorithm, [[Robotics|Robotics]], Autonomous Vehicle] last_reinforced: 2026-04-20 --- # [[Autonomous-Vehicle-Path-Planning|Autonomous-Vehicle-Path-Planning]] (μžμœ¨μ£Όν–‰ 경둜 κ³„νš) ## πŸ“Œ ν•œ 쀄 톡찰 (The Karpathy Summary) > 경둜 κ³„νšμ€ 'κ°€μž₯ λΉ λ₯Έ κΈΈ'을 μ°ΎλŠ” 것이 μ•„λ‹ˆλΌ, 'μ•ˆμ „ν•˜κ³  λΆ€λ“œλŸ¬μš°λ©° 예츑 κ°€λŠ₯ν•œ' μ›€μ§μž„μ„ μ‹€μ‹œκ°„μœΌλ‘œ μ„€κ³„ν•˜λŠ” ν™•λ₯ μ  탐색이닀. ## πŸ“– κ΅¬μ‘°ν™”λœ 지식 (Synthesized Content) - **Global Path Planning**: - μΆœλ°œμ§€μ—μ„œ λͺ©μ μ§€κΉŒμ§€μ˜ κ±°μ‹œμ μΈ 경둜λ₯Ό μ„€μ •ν•œλ‹€. 고전적인 **A* (A-star)** μ•Œκ³ λ¦¬μ¦˜μ΄λ‚˜ **Dijkstra** μ•Œκ³ λ¦¬μ¦˜μ΄ 지도 데이터 μœ„μ—μ„œ μž‘λ™ν•œλ‹€. - **Local Motion Planning (μ‹€μ‹œκ°„ νšŒν”Ό)**: - κ°‘μžκΈ° νŠ€μ–΄λ‚˜μ˜€λŠ” λ³΄ν–‰μžλ‚˜ μž₯애물을 ν”Όν•˜κΈ° μœ„ν•œ λ―Έμ‹œμ μΈ ꢀ적 μ΅œμ ν™”. **RRT* (Rapidly-exploring Random Tree)**λ‚˜ **Hybrid A*** 등이 μ‚¬μš©λœλ‹€. - **[[Behavior|Behavior]]al Decision (νŒλ‹¨ λ ˆμ΄μ–΄)**: - μ°¨μ„  λ³€κ²½, μΆ”μ›”, 일단 μ •μ§€ λ“± λ„λ‘œμ˜ λ²•κ·œ(Traffic Laws)와 에티켓을 λ°˜μ˜ν•œ μ˜μ‚¬κ²°μ • μ•Œκ³ λ¦¬μ¦˜κ³Ό 물리적 μ œμ–΄λ₯Ό κ²°ν•©ν•œλ‹€. ## ⚠️ λͺ¨μˆœ 및 μ—…λ°μ΄νŠΈ (RL Update) - κ³Όκ±°μ—λŠ” κ·œμΉ™ 기반(Rule-based) μ•Œκ³ λ¦¬μ¦˜μ΄ μ£Όλ₯˜μ˜€μœΌλ‚˜, μ΅œκ·Όμ—λŠ” λ³΅μž‘ν•œ 도심 상황을 ν•΄κ²°ν•˜κΈ° μœ„ν•΄ 'End-to-End' λ”₯λŸ¬λ‹ 방식과 '전톡적 ν”Œλž˜λ‹'을 κ²°ν•©ν•œ 계측적 ꡬ쑰가 ν‘œμ€€μœΌλ‘œ 자리 μž‘μ•˜λ‹€. ## πŸ”— 지식 μ—°κ²° (Graph) - Related: [[Systemic_Simulation_Principles|Systemic_Simulation_Principles]] , Robotic Manipulation - Context: [[Digital Twins|Digital Twins]]