Summit.OS - Distributed Intelligence Fabric

Summit.OS is Big Mountain Technologies' flagship command layer — an open, AI-driven "operating system for the physical world," designed to unify sensors, drones, and ground robots into a shared, real-time world model for decision-making and coordinated action.

🌍 System Concept

Summit.OS functions like Anduril's Lattice OS but is purpose-built for resilience, wildfire management, and critical-infrastructure protection. It sits above Linux, Mac, or Windows / ROS 2 and provides a unified intelligence fabric for data fusion, situational awareness, and autonomous tasking.

🧱 Core Architectural Layers

Layer	Description	Example Services
Edge Layer	Agents on robots / drones collecting telemetry, video, and sensor data; run local inference	/edge-agent, /ros-bridge, /edge-inference
Data Fabric Layer	Real-time message bus & synchronization (MQTT + gRPC + event mesh)	/fabric, /topics, /telemetry
Perception & Fusion Layer	Normalizes & fuses multi-modal data (video, weather, IR, lightning, soil) into a world model	/fusion, /classification, /tracking
Reasoning & Decision Layer	AI models generate contextual intelligence, predictions, & recommendations	/intelligence, /risk, /advisory
Command & Control Layer	Assigns missions & coordinates autonomous behaviors	/tasking, /autonomy, /swarm
Operator Interface	Real-time console for map, alerts, & mission management	/console (Next.js + MapLibre + 3D terrain)
Integration Layer	External systems (ArcGIS, CAD, dispatch, cloud APIs)	/integrations, /reports, /api

🧩 Technical Stack

• Backend: FastAPI (Python 3.11) + gRPC microservices
• Data Fabric: MQTT + Redis Streams + gRPC streaming
• Edge: ROS 2 / micro-ROS, ONNX Runtime, SQLite store-and-forward
• Frontend: Next.js 14 + TypeScript + MapLibre GL + shadcn/ui
• Database: Postgres + PostGIS + TimescaleDB
• AI/ML: PyTorch + Ray for distributed inference
• Infra: Docker Compose (dev), k3s (edge), EKS (cloud), Terraform IaC
• Auth & Security: Keycloak (OIDC RBAC), mTLS between agents
• Observability: OpenTelemetry + Prometheus + Grafana dashboards

🧠 AI Intelligence Architecture

Summit.OS is the distributed intelligence fabric that serves as the sense-making and autonomy brain for all BigMT robotics systems. AI is not an add-on feature—it's the connective tissue that enables every robot, drone, and sensor to understand, decide, and act in the physical world.

AI Integration by Layer

1. Data Fusion & Sensemaking (Perception Layer)

• Multimodal Sensor Fusion: Weather, LiDAR, IR, visual data → spatial context
• Anomaly Detection: Smoke, leaks, flooding, temperature spikes
• Terrain & Object Segmentation: Brush vs. road vs. water from drone imagery
• Environmental State Estimation: Dryness, wind vectors, fire spread rate

2. Object Detection, Classification & Tracking

• Computer Vision Models (ONNX/TensorRT): Detect smoke, tools, humans, vehicles
• Multisensor Tracking: Maintain persistent IDs across video frames and sensor types
• State Recognition: "valve open," "fireline constructed," "ditch cleared"
• Temporal Reasoning: Predict object motion or environmental changes

3. Autonomous Operations & Decision Reasoning

• Mission Planning: Graph-based planners + reinforcement learning for optimal task assignment
• Multi-agent Coordination: Decide which asset performs which subtask
• Predictive Modeling: Physics + ML hybrid models forecasting outcomes
• Contextual Advisory: Generate human-readable summaries for ops consoles

4. Learning & Continuous Improvement

• Model Retraining Pipelines: New sensor data improves detection accuracy
• Edge-to-Cloud Federated Learning: Local models train on field data, sync gradients
• Simulation-to-Real (Sim2Real): Reinforcement learning environments for robotics tasks
• Anomaly Feedback Loops: Operator confirmations fine-tune detection thresholds

Cross-Domain AI Applications

Product	Shared AI Services	Unique Additions
FireLine	Fusion, Intelligence, Tasking, Predict	Fire behavior, smoke detection
DitchBot	Fusion, Tasking, Predict	Water flow, soil erosion
OilfieldBot	Fusion, Tasking, Predict	Pressure anomalies, leak detection
GreaseBot	Fusion, Tasking	Fill-level estimation, route scheduling
TriageBot	Fusion, Intelligence, Tasking	Object/person recognition, triage prioritization

🎯 Functional Goals

1. Unified World Model – fuse all sensor & asset data into a live geospatial graph of fires, terrain, weather, and movement.
2. Autonomous Coordination – route multiple UGVs and UAVs dynamically; mission graphs (Patrol → Detect → Suppress → Verify).
3. Contextual Intelligence – produce human-readable situational alerts and recommendations.
4. Edge Resilience – operate offline with local buffering & sync on reconnect.
5. Operator Console – professional dark UI with map layers, device panels, and AI insight feed.
6. Open Architecture – allow third-party sensor / platform integration via standardized APIs.

🔗 FireLine Console Integration

FireLine Console (Next.js frontend) is the primary user-facing interface for the BigMT ecosystem. It connects to Summit.OS via API and real-time event channels to render live wildfire intelligence, mission status, and AI insights.

Integration Paths

Interface	Function	Example
REST/gRPC API Gateway	FireLine queries world model data, intelligence feeds, and mission status	GET /api/v1/intelligence/alerts, POST /api/v1/tasks
WebSocket / MQTT Streams	FireLine subscribes to live telemetry, detections, and status updates	ws://summit-os/fusion/stream or topic alerts/#
Event Hooks / Webhooks	Summit.OS pushes event-based updates to FireLine	POST /fireline/events/alert

🚀 Quick Start

# Clone and setup
git clone <repo-url>
cd Summit.OS

# Start development environment
make dev

# Access services
# - FireLine Console: http://localhost:3000
# - API Gateway: http://localhost:8000
# - Grafana: http://localhost:3001

📂 Monorepo Structure

/apps
  /console          # FireLine Console (Next.js frontend)
  /fabric           # Data fabric microservice
  /fusion           # Sensor fusion microservice
  /intelligence     # AI reasoning microservice
  /tasking          # Mission planning microservice
  /edge-agent       # Edge agent for robots/drones
  /integrations     # External system integrations
/packages
  /proto            # gRPC protocol definitions
  /schemas          # Shared data schemas
  /ui               # Shared UI components
  /geo              # Geospatial utilities
  /utils            # Common utilities
/infra
  /docker           # Docker configurations
  /k8s              # Kubernetes manifests
  /terraform        # Infrastructure as Code
/docs
  ARCHITECTURE.md
  API.md
  EDGE_PROTOCOL.md
/tests              # Integration and E2E tests

✅ Development Requirements

• Type-safe, modular, well-tested code
• Example data streams (robot telemetry + camera + weather)
• make dev spins up MQTT broker, Postgres, API gateway, and Console
• Unit tests for fusion, triangulation, and intelligence logic
• Mock mission demo: simulated fire detection → alert → automatic dispatch

🎯 Acceptance Criteria

1. make dev launches the full system locally with mock data
2. Console shows live telemetry and contextual AI alerts
3. Triangulation + fusion produce ignition estimates
4. Mission planner auto-tasks simulated UGVs & UAVs
5. All microservices documented and passing tests