Spanda

Bio-Inspired Resilient Autonomy Architecture

Canonical view: Cognitive & Resilience Architecture. This document is retained for backward compatibility and migration — see cognitive-resilience-migration.md.

Canonical functional view: cognitive-resilience-architecture.md — Spanda Cognitive & Resilience Architecture with eleven functional domains and a responsibility matrix. This document is retained for backward compatibility.

Spanda uses a hierarchical, resilient autonomy architecture inspired by proven principles from biological nervous systems: local reflexes, distributed control, sensory fusion, system homeostasis, adaptive recovery, platform immunity, and operational memory. The goal is not to imitate biology, but to apply these proven resilience patterns to safety-critical autonomous systems.

Caution: Spanda does not attempt to model consciousness, emotions, biological neurons, or artificial life. Biological metaphors are used only where they improve engineering resilience, safety, recovery, and explainability.

This architecture extends the existing Distributed Decision Architecture — it does not replace it.

Platform description

Spanda is a safety-first Autonomous Systems Platform with a dedicated programming language at its core. It provides readiness, assurance, diagnosis, recovery, trust, and distributed autonomy for robots, devices, AI agents, humans, and intelligent environments.

Architecture diagram

flowchart TB
  subgraph overlay [Cross-cutting resilience]
    R[Reflexes]
    A[Attention]
    H[Homeostasis]
    I[Immunity]
    M[Memory]
    RC[Recovery]
    T[Trust]
    AS[Assurance]
  end

  CC[Control Center / Cloud]
  REG[Regional Coordinator]
  FLEET[Fleet / Swarm Coordinator]
  ENT[Entity Runtime]
  REF[Reflex Controller]
  DEV[Sensors / Actuators / Devices]

  CC --> REG --> FLEET --> ENT --> REF --> DEV
  overlay -.-> CC
  overlay -.-> ENT
  overlay -.-> REF

Hierarchy (peripheral autonomy):

Control Center / Cloud
        ↓
Regional Coordinator
        ↓
Fleet / Swarm Coordinator
        ↓
Entity Runtime
        ↓
Reflex Controller
        ↓
Sensors / Actuators / Devices

Concepts

Concept	Purpose	Status	Guide
Reflex Arc	Immediate local safety response	Beta	reflex-architecture.md
Peripheral Autonomy	Avoid over-centralization	Beta	peripheral-autonomy.md
Sensory Fusion	Multi-source confidence	Experimental	sensory-fusion.md
Confidence Model	Conflict detection and fallback	Experimental	confidence-model.md
Attention System	Event prioritization	Beta	attention-system.md
Homeostasis	Safe operating range	Beta	homeostasis.md
Platform Immunity	Quarantine untrusted entities	Beta	platform-immunity.md
Operational Memory	Engineering memory categories	Preview	operational-memory.md
Habituation / Sensitization	Alert fatigue management	Experimental	habituation-sensitization.md
Damage-Risk Model	Harm potential, not just errors	Beta	damage-risk-model.md
Adaptive Recovery	Rule-based strategy learning	Experimental	adaptive-recovery.md
Maintenance / Sleep Mode	Low-risk maintenance windows	Preview	maintenance-mode.md

Entity integration

Every concept attaches to the unified Entity model via EntityAutonomyProfile:

Entity.reflexes — registered reflex summaries
Entity.attention — attention snapshot (priority, queue depth)
Entity.confidence — fused observation confidence
Entity.homeostasis — stability snapshot
Entity.immunity_status — quarantine state
Entity.memory_refs — operational memory references
Entity.damage_risk — harm-risk index
Entity.recovery_confidence — adaptive recovery score

REST: GET /v1/entities/{id}/autonomy

CLI

Command	Description
`spanda reflex list`	List platform reflex actions
`spanda reflex simulate [hint]`	Simulate reflex selection
`spanda reflex trace [hint]`	Emit reflex trace
`spanda fusion check`	Run sensory fusion demo
`spanda confidence report`	Confidence and conflict report
`spanda homeostasis check`	Evaluate stability metrics
`spanda homeostasis report`	Full homeostasis report
`spanda immunity scan`	Scan entities for immunity events
`spanda immunity quarantine [id]`	Quarantine decision for entity
`spanda alerts analyze`	Habituation / sensitization analysis
`spanda alerts fatigue-report`	Alert fatigue metrics
`spanda recovery confidence`	Recovery confidence from history
`spanda recovery learning-report`	Strategy preference report

gRPC

Autonomy REST routes mirror gRPC on Control Center (proto 1.0.14+): ListAutonomyReflexes, ListAutonomyReflexTraces, GetAutonomyHomeostasis, GetAutonomyImmunity, GetAutonomyAttention, GetAutonomyFusion, GetAutonomyMemory, GetEntityAutonomy.

Crate

Implementation lives in spanda-autonomy (crates/spanda-autonomy/).

Known limitations

See known-limitations.md. Sensory fusion remains rule-based; adaptive recovery uses statistics, not ML. Reflex traces persist under .spanda/autonomy-reflex-traces.json and are emitted during run/sim.

Smoke: ./scripts/cognitive_resilience_smoke.sh (CI Integration job bio-inspired-autonomy; alias for ./scripts/bio_inspired_autonomy_smoke.sh)

Language: homeostasis_policy and attention_policy declarations parse in .sd programs.

Distributed decisions — brain / spinal cord / reflex layers (Stable)
Platform architecture
Recovery orchestrator
Entity model

This site is open source. Improve this page.