Open 4D Spatiotemporal AI Platform built on PostgreSQL
📰 Featured: Why We Built Atlas4D — The problem with fragmented spatiotemporal data and how we solved it.
As seen in:
- Awesome-PostgreSQL - The definitive PostgreSQL resource list (Platforms category) 🆕
- Awesome-Geospatial - Curated list of geospatial resources
Atlas4D Base is the open-core of the larger Atlas4D platform. This repo contains a minimal but fully working 4D stack - database, core services, and observability. The full Atlas4D platform adds extra domain modules (radar, drones, telco network analytics, etc.).
- Data Engineers building real-time spatiotemporal pipelines
- GIS/Geo Developers needing time-series + vector search in one DB
- Telecom Teams monitoring network infrastructure
- Smart City Projects analyzing mobility and urban data
- Research Labs working with 4D trajectory data
| Feature | Traditional Approach | Atlas4D |
|---|---|---|
| Data Model | Separate geo, time, vector DBs | Unified PostgreSQL stack |
| Spatial Indexing | R-tree only | H3 hexagons + PostGIS |
| Time Series | Separate TSDB | TimescaleDB integrated |
| Vector Search | External service | pgvector in-database |
| ML Pipeline | Build from scratch | Ready anomaly/threat detection |
| Natural Language | Not included | NLQ query interface |
| Observability | DIY | Prometheus + Grafana + Alerts |
# Clone and start
git clone https://github.com/crisbez/atlas4d-base.git
cd atlas4d-base
make demoOr manually:
cp .env.example .env
docker compose up -d --build⏱️ Time to first map: ~3 minutes
| Service | URL |
|---|---|
| Map UI | http://localhost:8091 |
| API Health | http://localhost:8090/health |
| API Stats | http://localhost:8090/api/stats |
Try Atlas4D without installing:
| Service | URL |
|---|---|
| Map UI | atlas4d-demo.digicom.bg:8091 |
| API Health | /health |
| API Stats | /api/stats |
| Observations | /api/observations |
💡 Demo resets daily with fresh Burgas mobility data.
Atlas4D is built as a set of independent services around a shared 4D database core.
- Add new domain modules without touching the core DB
- Mix & match services (anomaly only, or anomaly + threat + NLQ)
- Safe to extend: everything talks HTTP/JSON or SQL
- Scalable: suitable for single-node labs and multi-service production clusters
┌─────────────────────────────────────────────────────────────┐
│ Atlas4D Platform │
├─────────────────────────────────────────────────────────────┤
│ ┌─────────┐ ┌─────────┐ ┌─────────┐ ┌─────────┐ │
│ │ Map │ │ NLQ │ │ Health │ │ API │ UI │
│ │ UI │ │ Chat │ │Dashboard│ │ Docs │ │
│ └────┬────┘ └────┬────┘ └────┬────┘ └────┬────┘ │
│ │ │ │ │ │
│ ┌────┴────────────┴────────────┴────────────┴────┐ │
│ │ API Gateway (FastAPI) │ │
│ └────┬────────────┬────────────┬────────────┬────┘ │
│ │ │ │ │ │
│ ┌────┴────┐ ┌────┴────┐ ┌────┴────┐ ┌────┴────┐ │
│ │ Anomaly │ │ Threat │ │Embedding│ │ Public │Services│
│ │ Svc │ │Forecast │ │ Svc │ │ API │ │
│ └────┬────┘ └────┬────┘ └────┬────┘ └────┬────┘ │
│ │ │ │ │ │
│ ┌────┴────────────┴────────────┴────────────┴────┐ │
│ │ PostgreSQL + PostGIS + TimescaleDB │ Data │
│ │ + H3 + pgvector │ Layer │
│ └─────────────────────────────────────────────────┘ │
└─────────────────────────────────────────────────────────────┘
- PostgreSQL 16 - Rock-solid foundation
- PostGIS 3.4 - Spatial operations and geometry
- TimescaleDB - Time-series hypertables with compression
- H3 - Uber's hexagonal hierarchical indexing
- pgvector - Vector similarity search for embeddings
- public-api - REST API for data ingestion and queries
- anomaly-svc - Real-time anomaly detection (reference models)
- threat-forecastor - ML-powered threat assessment (reference model)
- trajectory-embedding - Trajectory vectorization with caching
- nlq-svc - Natural language to SQL translation (Bulgarian + English)
- Prometheus - Metrics collection
- Grafana - Dashboards and visualization
- Alert Rules - Pre-configured for ML pipeline and Redis
| Aspect | Status |
|---|---|
| Maturity | Tech Preview / Alpha |
| Scope | Core 4D database, reference AI services, observability |
| Not included | Domain-specific modules (radar, ADS-B, vision GPU, NetGuardian) - see Full Edition |
Atlas4D Base ships with the core 4D engine and generic AI services. On top of this core, domain-specific modules can be added:
- GPON / fiber anomaly detection
- Capacity & congestion forecasting
- Network Guardian-style risk scoring for critical infrastructure
- Traffic & fleet analytics via trajectories
- Movement anomalies (speed spikes, unusual routes)
- High-risk zones (stadiums, events, gatherings)
- Trajectory monitoring for aircraft and drones
- Conflict zones / separation violation detection
- Safety dashboards for control rooms
- Fire risk mapping (wind, temperature, drought, historical fires)
- Crop yield forecasting on H3 grid
- Early warning for extreme weather events
- Multi-sensor drone detection (radar + vision + RF)
- Spatiotemporal analysis of suspicious objects and vehicles
- Pattern-of-life analysis on 4D trajectories
- Time-series forecasting
- Vector-based similarity: "find trajectories similar to this incident"
Atlas4D Base ships with a developer-friendly demo configuration.
It is not production-ready out of the box.
Before exposing a deployment to the internet you should:
- change all default passwords and secrets,
- restrict exposed ports and put Atlas4D behind a reverse proxy (HTTPS),
- use a dedicated DB user with least privilege,
- protect observability (Grafana/Prometheus) and internal APIs.
See docs/SECURITY.md for a detailed hardening guide.
- Quick Start Guide
- Architecture Overview
- Database Schema
- API Reference
- NLQ Usage Guide
- STSQL Overview
# .env.example
POSTGRES_HOST=postgres
POSTGRES_DB=atlas4d
POSTGRES_USER=atlas4d_app
POSTGRES_PASSWORD=your_secure_password
# Optional: Enable ML features
ENABLE_ANOMALY_DETECTION=true
ENABLE_THREAT_FORECAST=true
ENABLE_NLQ=true- Core spatiotemporal database schema
- H3 hexagonal indexing
- Anomaly detection pipeline
- Embedding cache with Redis
- Natural language queries (Bulgarian + English)
- E2E demo test suite
- Kubernetes Helm charts
- Multi-tenant support
- Real-time WebSocket feeds
This is Atlas4D Base - the open-source foundation and reference implementation.
Atlas4D Full Edition includes additional enterprise modules built on top of this base:
- Radar & ADS-B fusion for airspace monitoring
- Drone & low-altitude threat detection
- Telco Network Guardian (fiber/ISP network analytics)
- GPU-accelerated vision and video analytics
- Advanced forecasting (multi-source risk scoring, LSTM models)
- SLA-grade support, sizing and deployment guidance
Contact us for enterprise inquiries.
Atlas4D is designed for multiple industries. Explore our use case outlines:
| Industry | Use Case | Key Features |
|---|---|---|
| 📡 Telecom | Network Monitoring | SNMP, GPON, real-time alerts |
| 🏙️ Smart City | Municipal IoT Platform | Traffic, parking, air quality |
| 🛡️ Defense | Counter-Drone System | Radar fusion, swarm detection |
| 🔥 Emergency | Wildfire Monitoring | Risk prediction, satellite hotspots |
| 🌾 Agriculture | Harvest Forecasting | NDVI tracking, yield prediction |
Each case study includes architecture diagrams, NLQ examples (Bulgarian + English), and expected results.
We welcome contributions! See our Contributing Guide.
Ways to contribute:
- 🐛 Report bugs
- 📝 Improve documentation
- 💻 Submit code
- 💡 Propose new modules
Check out our Good First Issues for beginner-friendly tasks.
Apache 2.0 - See LICENSE for details.
Atlas4D stands on the shoulders of giants:
Our vision: Atlas4D aims to become the "Linux of 4D spatiotemporal data platforms" - a stable, open foundation for location-aware, time-sensitive AI applications.
⭐ Star this repo if you find it useful!
Atlas4D supports natural language queries in Bulgarian and English:
Bulgarian:
- "Какво е времето в Бургас?"
- "Покажи заплахи около София"
- "Покажи аномалии от последния час"
English:
- "Show threats near the airport"
- "What anomalies happened today?"
See NLQ Usage Guide for full examples.
| Service | URL |
|---|---|
| Map UI | http://185.18.56.13:8091 |
| API | http://185.18.56.13:8090/api/observations |
| Health | http://185.18.56.13:8090/health |
| Stats | http://185.18.56.13:8090/api/stats |
New to Atlas4D Base? Start here:
- Developer Onboarding - Architecture, first 10 minutes, common tasks
- Contributing Guide - How to submit PRs, code style, testing
# Start stack
docker compose up -d
# View logs
docker compose logs -f api-gateway
# Connect to database
docker compose exec postgres psql -U atlas4d_app -d atlas4d
# Rebuild after changes
docker compose build api-gateway && docker compose up -d api-gatewaySee ROADMAP.md for upcoming features and releases.
Current: v0.3.0 | Next: v0.4.0 (Q1 2026) — Module Ecosystem