2nd/10_Wiki/Topics/AI_and_ML/CPTED.md

id, title, category, status, canonical_id, aliases, duplicate_of, source_trust_level, confidence_score, verification_status, tags, raw_sources, last_reinforced, github_commit, tech_stack

id	title	category	status	canonical_id	aliases	duplicate_of	source_trust_level	confidence_score	verification_status	tags	raw_sources	last_reinforced	github_commit	tech_stack
wiki-2026-0508-cpted	CPTED (Crime Prevention Through Environmental Design)	10_Wiki/Topics	verified	self	CPTED environmental design defensible space broken windows urban safety Oscar Newman	none	A	0.88	applied	urban-planning security cpted defensible-space broken-windows smart-city environmental-design		2026-05-10	pending	language applicable_to urban planning / security Urban Design Smart City Security Architecture

CPTED (Crime Prevention Through Environmental Design)

📌 한 줄 통찰

"매 공간 으로 범죄 의 prevention". 매 CCTV 의 reactive X — 매 building / lighting / fence 의 design 의 의지 의 deter. 매 5 strategy. 매 modern: 매 smart city + AI 의 simulation 의 augment.

📖 핵심

매 5 strategy (2nd-gen CPTED)

1. Natural Surveillance: 매 visibility (low fence, transparent wall, lighting).
2. Natural Access Control: 매 single entry, 매 clear pathway.
3. Territorial Reinforcement: 매 public-private boundary 의 clear.
4. Activity Support: 매 people 의 traffic.
5. Maintenance (Image): 매 broken window 의 fix.

매 historical

• Jane Jacobs (1961): "The Death and Life of Great American Cities" — 매 "eyes on the street".
• Oscar Newman (1972): "Defensible Space".
• C. Ray Jeffery (1971): 매 CPTED term.
• Wilson & Kelling (1982): 매 broken windows theory.

매 1st vs 2nd vs 3rd generation

세대	강조
1st	매 physical (Jeffery, Newman)
2nd	매 social cohesion (community)
3rd	매 sustainability + tech

매 application 예

• Park: 매 sightline + lighting + 매 wide path.
• Apartment: 매 lobby visibility + 매 single entry + 매 maintained.
• Parking lot: 매 lighting + 매 emergency phone + 매 cctv.
• School: 매 layered security + 매 visibility + 매 community.
• ATM: 매 lighting + 매 visibility + 매 mirror.
• Transit station: 매 sightline + 매 staff presence.

매 modern (smart city)

• AI surveillance: 매 abnormal pattern detection.
• Adaptive lighting: 매 motion-triggered.
• Crowd flow analytics: 매 design feedback.
• Predictive crime mapping: 매 high-risk area focus.
• Citizen reporting app: 매 311 / SeeClickFix.

매 limitation / critique

• Displacement: 매 crime 의 다른 area 의 move.
• Surveillance: 매 privacy concern.
• Equity: 매 wealthy area 의 over-invest.
• False sense: 매 design 의 omnipotent X.
• Broken windows critique: 매 racial bias.

매 Korea CPTED

• 매 2014 의 시범 도시.
• 매 경찰청 의 cooperation.
• 매 Salt Path / 안심 귀가 길.
• 매 mural / lighting / mirror.

매 design checklist

1. 매 sightline 의 unobstructed?
2. 매 lighting 의 0.5+ lux 의 every spot?
3. 매 access route 의 single + clear?
4. 매 dead-end / hidden alcove?
5. 매 maintenance 의 < 24h response?
6. 매 territoriality (sign, paint, fence)?
7. 매 activity (cafe, store) 의 generator?

💻 패턴 (응용 — design checklist + sim)

CPTED audit checklist (programmatic)

def cpted_audit(location):
    return {
        'natural_surveillance': {
            'sightline_coverage': measure_sightlines(location),  # 매 % visible
            'avg_lux_at_night': measure_lighting(location),
            'window_facing_ratio': building_facade_ratio(location),
        },
        'access_control': {
            'entry_count': count_entries(location),
            'pathway_clarity': measure_path_clarity(location),
        },
        'territoriality': {
            'boundary_markers': count_boundary_signs(location),
            'private_public_clarity': assess_boundary(location),
        },
        'activity': {
            'foot_traffic_per_hour': pedestrian_count(location),
            'commercial_density': commerce_per_sqm(location),
        },
        'maintenance': {
            'graffiti_density': count_graffiti(location),
            'broken_lighting_pct': pct_broken_lights(location),
            'litter_score': litter_density(location),
        },
    }

Crime risk simulation

def predict_crime_risk(area, design_params):
    """매 simple model 의 risk score."""
    risk = 0
    risk -= design_params['lux_avg'] * 0.3
    risk -= design_params['sightline_pct'] * 0.5
    risk += design_params['hidden_alcoves'] * 2
    risk -= design_params['foot_traffic_per_hr'] * 0.01
    risk += design_params['litter_score'] * 0.5
    return max(0, risk)

# 매 design alternative 의 비교
baseline = predict_crime_risk(area, current_design)
improved = predict_crime_risk(area, {**current_design, 'lux_avg': 5, 'hidden_alcoves': 0})
print(f'Risk reduction: {baseline - improved:.1f}')

Adaptive lighting (smart city)

class AdaptiveStreetlight:
    def __init__(self, motion_sensor, schedule):
        self.sensor = motion_sensor
        self.schedule = schedule
    
    def update(self):
        time = datetime.now().time()
        # 매 base level
        base_level = self.schedule.level_for(time)
        
        # 매 motion 시 의 brighten
        if self.sensor.motion_detected_recently(seconds=30):
            self.set_brightness(min(100, base_level + 50))
        else:
            self.set_brightness(base_level)

311 / citizen report integration

def cpted_response_pipeline(report):
    """매 citizen report → 매 prioritize."""
    if report.type == 'broken_streetlight':
        priority = 'high' if report.area.crime_rate > MEDIAN else 'medium'
        target_response = 24 if priority == 'high' else 72  # hours
    elif report.type == 'graffiti':
        priority = 'medium'
        target_response = 48
    elif report.type == 'overgrown_bush':
        priority = 'medium'  # 매 sightline 의 obstruct
        target_response = 72
    
    return dispatch(report, priority, target_response)

Design alternative scorer

def score_design_options(options):
    scored = []
    for opt in options:
        score = (
            opt.surveillance_score * 0.3 +
            opt.access_control_score * 0.2 +
            opt.territoriality_score * 0.2 +
            opt.activity_score * 0.2 +
            opt.maintenance_score * 0.1
        )
        cost = opt.estimated_cost
        scored.append((opt, score, score / cost))  # 매 cost-effectiveness
    return sorted(scored, key=lambda x: -x[2])

Predictive crime mapping (caution)

# 매 ProPublica / Gender Shades 의 lesson:
# 매 historical crime data 의 bias.
# 매 over-policing 의 reinforce.

def predict_with_bias_check(features, model, bias_audit):
    pred = model.predict(features)
    
    # 매 demographic 의 audit
    by_demo = bias_audit.check(pred)
    if by_demo['disparity'] > 0.2:
        flag('Disparate impact detected — review required')
    
    return pred

🤔 결정 기준

상황	Strategy
Park redesign	Sightline + lighting + activity
Apartment	Single entry + lobby visibility
Parking	Lighting + emergency call + visibility
Transit	Sightline + staff + cctv
Smart city	Adaptive lighting + crowd analytics
Tight budget	Lighting + maintenance

기본값: 매 surveillance + lighting + maintenance + activity 의 first investment.

🔗 Graph

• 부모: Security
• 변형: Defensible-Space · Broken-Windows
• 사상가: Oscar-Newman
• Adjacent: Atmospheric-Intelligence · Algorithmic Fairness

🤖 LLM 활용

언제: 매 urban planning. 매 building design. 매 smart city. 매 community safety initiative. 언제 X: 매 systemic root cause (poverty 의 substitute). 매 surveillance state justification.

❌ 안티패턴

• Surveillance 의 only: 매 design 의 ignore.
• Fortress design: 매 community 의 disconnect.
• No maintenance: 매 broken windows.
• No activity: 매 dead street.
• Bias 의 ignore (predictive crime): 매 over-policing.
• Rich neighborhoods 만 의 invest: 매 inequity.

🧪 검증 / 중복

• Verified (Jacobs 1961, Newman 1972, Wilson-Kelling 1982).
• 신뢰도 A.
• Related: Smart-City · Atmospheric-Intelligence · Surveillance-Capitalism · Algorithmic Fairness.

🕓 Changelog

날짜	변경
2026-05-08	Phase 1
2026-05-10	Manual cleanup — 5 strategy + history + smart-city + 매 audit / sim / adaptive code