Skip to content

platform-overview.md

Latest commit

 

History

History
282 lines (213 loc) · 13.5 KB

platform-overview.md

File metadata and controls

282 lines (213 loc) · 13.5 KB
Raw

Spanda Platform Overview

Spanda logo

Spanda is a safety-first Autonomous Systems Platform with a dedicated programming language at its core.

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.

Short form: The Autonomous Systems Platform.

Pronounced SPUN-duh (/ˈspʌndə/) — Sanskrit for the divine pulse; see overview/philosophy.md.

This document explains how the Spanda Platform relates to the Spanda Language, what each major component does, and where to go next.

Platform vs language

Spanda Platform Spanda Language
What it is End-to-end toolchain for designing, verifying, simulating, deploying, and operating autonomous systems The .sd programming language — one core component of the platform
Scope Runtime, verification, safety engine, simulation, replay, health, fleet, packages, providers, tooling Syntax, types, robot primitives, safety types, units, and compile-time checks
How you use it spanda check, verify, sim, replay, fleet, install, demo, … Write .sd source; import packages; declare robot, sensor, actuator, safety, deploy
Identity Build. Verify. Simulate. Deploy. Operate. Robot-native, safety-typed, unit-aware source code

The language is not being replaced or renamed. .sd files, language tutorials, and language reference docs remain first-class. The platform framing adds context: Spanda is more than a compiler — it is the layer where autonomous systems are designed, validated, and operated.

Previous positioning: The Autonomous Systems Language — still accurate for the language itself; the platform name reflects the full product scope. See platform-positioning-migration.md for messaging guidance.

Spanda Platform architecture

Spanda Platform
│
├── Spanda Language (.sd)
├── Spanda Runtime
├── Spanda Verify
├── Spanda Safety
├── Spanda Sim
├── Spanda Replay
├── Spanda Health
├── Spanda Readiness
├── Spanda Mission Assurance
├── Spanda Fleet
├── Spanda Registry
├── Spanda Providers
└── Enterprise Operations (Control Center, Device Pool, APIs, …)

Cognitive & Resilience Architecture

Spanda implements a Cognitive & Resilience Architecture inspired by proven engineering principles observed in biological nervous systems. Rather than replicating biological anatomy, it 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. This extends the Distributed Decision Architecture without replacing it.

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.

Guides: cognitive-resilience-architecture.md · functional-domains.md · responsibility-matrix.md

Spanda Language (.sd)

The safety-first programming language at the center of the platform.

  • Source files use the .sd extension (unchanged).
  • First-class concepts: robot, sensor, actuator, ai_model, agent, safety, task, deploy, hardware, verify { }, physical units.
  • Safety types: ActionProposal (untrusted AI output) vs SafeAction (post-safety.validate()).
  • Reference: spanda-language.md, spanda-reference.md, tutorials/README.md.

Spanda Runtime

Execution layer after compile-time and certification gates.

  • Tree-walking interpreter (spanda-interpreter) as the primary execution path.
  • Scheduler, cooperative concurrency (spawn, join, select), HAL bindings, trigger-driven execution.
  • Provider dispatch connects official packages to live or mock backends.
  • Crates: spanda-runtime, spanda-interpreter, spanda-concurrency, spanda-hal, spanda-certify.
  • Reference: architecture.md, concurrency.md, triggers.md.

Spanda Verify

Pre-deploy and continuous verification.

  • Hardware verification: spanda verify — memory, sensors, actuators, timing, battery, network, AI model requirements against hardware profiles.
  • Capability verification: exposure, grants, traceability matrices, minimum-hardware analysis.
  • Behavioral verification: verify { } blocks and simulate_compatibility fault injection.
  • Reference: hardware-compatibility.md, capability-traceability.md, ci-verify.md.

Spanda Safety

Safety engine integrated with the type system and runtime.

  • Compile-time gate: actuators require SafeAction, not ActionProposal.
  • Runtime rules: max_speed, safety zones, stop_if, emergency stop.
  • Kill switch: kill_switch, remote_signed, on kill_switch handlers.
  • Reference: agentic-programming.md, kill-switch.md.

Spanda Sim

Simulation without physical hardware.

  • spanda run and spanda sim with physics-lite 2D backend.
  • Digital twins: twin { mirror …; replay true; } for shadow state.
  • Fault injection via simulate_compatibility.
  • Reference: killer-demo.md, examples in examples/showcase/.

Spanda Replay

Mission trace capture and playback.

  • Record: spanda sim --record → mission trace files.
  • Replay: spanda replay --deterministic (parity check) or --playback (state snapshots).
  • Reference: replay.md, realtime.md.

Spanda Health

Operational health monitoring and fleet readiness.

  • health_check, health_policy, fleet require clauses evaluated at runtime.
  • Integration with verification diagnostics and capability analysis.
  • Reference: health-checks.md, fleet-health.md.

Spanda Readiness

Weighted operational go/no-go scoring before deploy and during operations.

  • spanda readiness — composite score across verification, health, assurance, and deploy fit.
  • Agent APIs: spanda deploy agent readiness, fleet readiness aggregates.
  • Reference: readiness.md, fleet-readiness.md.

Spanda Mission Assurance

Mission-grade autonomous operations assurance integrated with readiness and verification.

Spanda Fleet

Multi-robot coordination and distributed operation.

  • In-process: spanda fleet run for multi-robot simulation.
  • Orchestration: spanda fleet orchestrate, mesh coordinator, HTTP agent relay (--remote, --mesh-url).
  • OTA: deploy plan, rollout, rollback, remote agents.
  • Reference: concurrency.md, fleet-distributed.md.

Spanda Registry

Package distribution and discovery.

  • Hosted index (registry/index.json) with Ed25519-signed tarballs.
  • CLI: spanda install, spanda update, spanda publish.
  • Override registry URL via SPANDA_REGISTRY_URL.
  • Reference: registry.md, packages.md.

Spanda Providers

Extensibility through official and community packages.

Enterprise Operations (Control Center)

Production operations layer for fleet visibility, provisioning, governance, and integration — composes existing engines without duplicating them.

  • Control Center: spanda control-center serve — React/TypeScript UI (ControlCenterPanel, sidebar version), Rust spanda-api backend, Tauri desktop 0.6.3 (desktop-v0.6.3 GitHub Release). Versioning: control-center-versioning.md.
  • Device Pool & Provisioning: Central inventory, lifecycle states, discover → verify → assign → ready workflow.
  • Governance: RBAC, secret management, alerting, compliance exports, digital thread query.
  • Integration: REST v1, Python SDK, WebSocket telemetry, OTLP observability.
  • Reference: enterprise-operations-roadmap.md, control-center.md.

Typical workflow

Write (.sd)  →  check  →  verify  →  sim  →  deploy  →  operate
     │            │          │         │        │          │
 Language    type/units   hardware   safety   target    health
             safety       capability  replay   fleet    readiness
                                       assurance replay
                                       continuity takeover
  1. Build — Author robot programs in .sd with safety rules and deployment targets.
  2. Verify — spanda verify against hardware profiles; capability and behavioral checks.
  3. Simulate — spanda sim before hardware is available; inject faults.
  4. Assure — spanda assure / anomaly scan / state estimate for mission assurance reports (optional spanda demo assurance).
  5. Deploy — deploy Robot to Profile; optional native codegen (experimental) or interpreter on edge.
  6. Operate — Health policies, readiness scoring, fleet coordination, mission continuity (takeover/delegation/succession), self-healing recovery, and mission replay for incidents.
  7. Govern — Control Center for fleet visibility; RBAC, secrets, alerting, compliance exports (experimental).

Flagship walkthrough: killer-demo.md · Platform demo: examples/showcase/autonomous_rover/README.md · Mission assurance: examples/showcase/assurance/README.md · Mission continuity: examples/showcase/continuity/README.md.

What Spanda is not

Spanda does not replace Python for ML training, C++ for low-level drivers, ROS2 for production transport at scale, or Gazebo/Isaac for full physics simulation. It coordinates and verifies autonomous logic across those layers — with safety and deploy fit enforced in one platform.

See product-strategy.md for competitive positioning and feature-status.md for honest capability tiers.

Related documents

Document Purpose
README.md Project home, quick start, examples
vision.md Long-term vision and philosophy
product-strategy.md Priorities, pillars, release scope
roadmap.md Roadmap by platform area
enterprise-operations-roadmap.md Control Center, Device Pool, provisioning, APIs (20 pillars)
control-center.md spanda control-center serve, REST v1, desktop 0.6.3 release · control-center-versioning.md
desktop-release-runbook.md Tauri desktop desktop-v* tags, GitHub Releases, optional signing
mission-assurance.md Mission assurance CLI, packages, and examples
mission-continuity.md Mission continuity, takeover, delegation, succession
continuity-policies.md continuity_policy syntax and validation
platform-positioning-migration.md Messaging migration notes
architecture.md Compiler pipeline and crate map
lean-core.md Lean-core workspace design