Spanda

Spanda Cognitive & Resilience Architecture

Spanda implements a Cognitive & Resilience Architecture — a functional view of platform responsibilities inspired by proven engineering principles observed in biological nervous systems. Rather than replicating biological anatomy, Spanda adopts functional concepts such as local reflexes, distributed coordination, sensory fusion, homeostasis, platform immunity, operational memory, adaptive recovery, and attention management to improve safety, resilience, and explainability.

Important: Spanda is not a biologically inspired AI platform and does not attempt to model consciousness, emotions, or neural structures. Biological concepts are used only where they provide measurable engineering benefits. Implementation modules are not named after brain anatomy (no Cortex, Cerebellum, Hippocampus, etc.).

This architecture sits alongside existing views — it does not replace them:

Architectural view Focus
Platform Architecture Layered crate structure, dependency governance
Entity Architecture Unified Entity model and graph
Runtime Architecture Compiler pipeline, interpreter, scheduler
Distributed Decision Architecture Reflex / local / fleet / control-center layers
Recovery Architecture Plan, simulate, execute, audit recovery
Cognitive & Resilience Architecture (this document) Functional responsibility domains

Implementation crate: spanda-autonomy — wraps and integrates existing platform services; does not duplicate them.

Platform description

Spanda is a safety-first Autonomous Systems Platform with a dedicated programming language at its core. It orchestrates robots, devices, AI agents, vehicles, humans, and intelligent environments using a unified Entity Model and built-in capabilities for readiness, assurance, recovery, trust, health, distributed autonomy, and governance.

Architecture diagram

flowchart TB
  subgraph strategic [Strategic Planning]
    MP[Mission Planner]
    GOV[Governance]
    SIM[Simulation]
  end

  subgraph operational [Operational Coordination]
    FLEET[Fleet]
    MR[Mission Runtime]
    DD[Distributed Decisions]
  end

  subgraph reflex [Reflex and Safety]
    REF[Reflex Controller]
    KS[Kill Switch]
    RT[Runtime Protection]
  end

  subgraph homeo [Homeostasis Engine]
    HL[Health]
    TEL[Telemetry]
    RD[Readiness]
  end

  subgraph immune [Platform Immunity]
    SEC[Security]
    TR[Trust]
    TAM[Tamper Detection]
  end

  subgraph fusion [Sensory Fusion]
    CONF[Confidence Model]
  end

  subgraph attention [Attention Engine]
    ALT[Alerts]
    TEL2[Telemetry]
  end

  subgraph memory [Operational Memory]
    RPL[Replay]
    KG[Knowledge Graph]
    PB[Playbooks]
  end

  subgraph adaptive [Adaptive Learning]
    STATS[Statistics]
    RULES[Rule-based Adaptation]
  end

  subgraph risk [Damage Risk Assessment]
    DR[Damage Risk Model]
  end

  subgraph maint [Maintenance and Optimization]
    OTA[OTA]
    CAL[Calibration]
  end

  subgraph recovery [Recovery]
    RO[Recovery Orchestrator]
  end

  ENT[Entity Model]

  strategic --> operational
  operational --> reflex
  homeo --> operational
  immune --> operational
  fusion --> homeo
  attention --> operational
  memory --> strategic
  adaptive --> recovery
  risk --> strategic
  maint --> homeo
  recovery --> operational

  ENT -.-> strategic
  ENT -.-> operational
  ENT -.-> reflex
  ENT -.-> homeo
  ENT -.-> immune
  ENT -.-> fusion
  ENT -.-> attention
  ENT -.-> memory
  ENT -.-> adaptive
  ENT -.-> risk
  ENT -.-> recovery

Peripheral autonomy hierarchy (unchanged from distributed decisions):

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

Functional domains

Eleven functional responsibility domains organize platform capabilities. Full definitions: functional-domains.md.

Domain Primary responsibility Status
Strategic Planning Mission planning, policy, governance, deployment Stable (mission planner, governance)
Operational Coordination Fleet, scheduling, delegation, takeover Stable
Reflex & Safety Emergency stop, kill switch, local safety Beta
Homeostasis Engine Stable operating conditions Beta
Platform Immunity Quarantine, isolation, re-admission Beta
Sensory Fusion Multi-source confidence Beta (entity-derived rule-based fusion)
Attention Engine Event prioritization, alert fatigue Beta
Operational Memory Mission state, replay, playbooks Beta
Adaptive Learning Rule-based outcome learning Experimental
Damage Risk Assessment Harm potential modeling Beta
Maintenance & Optimization OTA, calibration, maintenance windows Stable (OTA) / Beta (windows)

Entity integration

Every functional domain operates on Entity objects. Domain state attaches via Entity.autonomy (EntityAutonomyProfile):

Entity field Functional domain
Entity.health Homeostasis Engine (via health_status)
Entity.homeostasis Homeostasis Engine
Entity.readiness Homeostasis Engine, Sensory Fusion
Entity.trust Platform Immunity
Entity.immunity (immunity_status) Platform Immunity
Entity.confidence Sensory Fusion
Entity.reflexes Reflex & Safety
Entity.attention Attention Engine
Entity.operationalMemory (memory_refs) Operational Memory
Entity.damageRisk (damage_risk) Damage Risk Assessment
Entity.recoveryConfidence Adaptive Learning, Recovery

REST: GET /v1/entities/{id}/autonomy — enriched profile with all domain snapshots.

Responsibility matrix

Capability-to-domain mapping prevents service overlap: responsibility-matrix.md.

SDK and API

Domain clients wrap existing REST endpoints — no duplicate business logic:

Client REST prefix Guide
ReflexClient /v1/autonomy/reflex reflex-architecture.md
HomeostasisClient /v1/autonomy/homeostasis platform-homeostasis.md
ImmunityClient /v1/autonomy/immunity platform-immunity.md
AttentionClient /v1/autonomy/attention attention-engine.md
FusionClient /v1/autonomy/fusion sensory-fusion.md
MemoryClient /v1/autonomy/memory operational-memory.md
RiskClient /v1/risk, entity damage_risk damage-risk.md

AutonomyClient remains as a backward-compatible facade. See entity-sdk.md.

gRPC parity (proto 1.0.14+): ListAutonomyReflexes, GetAutonomyHomeostasis, GetAutonomyImmunity, GetAutonomyAttention, GetAutonomyFusion, GetAutonomyMemory, GetEntityAutonomy.

Control Center

The Cognitive & Resilience tab organizes live panels by functional domain: Strategic Planning summary, Reflex Events, Attention Queue, Homeostasis, Platform Immunity, Operational Memory, Damage Risk, Recovery Confidence. See control-center.md.

CLI

Command Domain
spanda reflex list\|simulate\|trace Reflex & Safety
spanda homeostasis check\|report Homeostasis Engine
spanda immunity scan\|quarantine\|report Platform Immunity
spanda fusion check Sensory Fusion
spanda confidence report Sensory Fusion
spanda alerts analyze\|fatigue-report Attention Engine
spanda recovery confidence\|learning-report Adaptive Learning

Migration

Existing bio-inspired-architecture.md content is preserved. The Cognitive & Resilience Architecture is the canonical functional view going forward. See cognitive-resilience-migration.md.

This site is open source. Improve this page.