Spanda is an Autonomous Systems Platform with a safety-first programming language at its core. The pulse of autonomous intelligence.
The Spanda Language (.sd) remains the expressive center — robot-native syntax, safety types,
and physical units. The platform adds verification, simulation, replay, health, fleet, packages, and
providers around that core.
Platform guide: platform-overview.md
Hardware is the body.
Sensors are the senses.
AI models are the mind.
Actuators are the muscles.
Spanda is the intelligent pulse that transforms perception, intent, and safety into action.
Autonomy is layered: reflex for safety, spinal cord for bounded local action, brain for strategy — with AI as mind, not master. Expanded philosophy
Cognitive & Resilience Architecture extends that hierarchy with homeostasis, platform immunity, sensory fusion, operational memory, and adaptive recovery — cognitive-resilience-architecture.md.
Spanda (Pronounced SPUN-duh (/ˈspʌndə/)) is a Sanskrit term meaning the divine pulse — the creative vibration of consciousness and energy that manifests as expansion and contraction in all entities, bridging stillness and movement within consciousness; and the first stir of awareness that creates and sustains the universe.
Expanded philosophy: overview/philosophy.md
Autonomous systems — robots, drones, industrial arms, warehouse fleets, humanoid assistants — are built from fragments:
No single language treats perception, intelligence, safety, verification, and deployment as unified first-class concepts. Teams spend more time integrating than innovating.
Spanda is the platform where autonomous systems are designed, validated, simulated, deployed, and
operated — with the .sd language as the core and AI treated as untrusted input, hardware fit
checked before code ships.
Sensors → Perception → AI Planning → Safety Gate → Actuators
↑ ↑
Digital Twin Hardware Verify
We believe the next generation of robotics and autonomous agents needs a platform that speaks their domain natively — with a dedicated language at the center, not a general-purpose language with libraries bolted on.
AI models are advisors, not drivers. LLMs and vision models propose ActionProposal values;
only safety.validate() produces SafeAction that may reach actuators. Agents have goals, memory,
and tools — but always within safety and capability constraints.
Sensors, actuators, trajectories, safety zones, and deterministic task scheduling are language
primitives — not framework imports. Physical units (m/s, rad, m) are enforced at compile time.
Every physical robot can have a twin that mirrors pose, velocity, and sensor state. Replay buffers
enable post-incident analysis and simulation sync. Live cloud twin telemetry is on the roadmap.
Safety is not optional middleware. It is woven into the type system (ActionProposal vs
SafeAction), runtime monitor (stop_if, zones, emergency stop), and verification (verify { }
behavioral assertions).
Before deployment, Spanda answers: does this program fit this hardware? Memory, sensors, timing,
power, network, and AI model requirements are checked against hardware profiles via spanda verify.
Agents can interact with humans through structured communication primitives — topics, services, actions, and event handlers — with audit trails for accountability.
| Spanda is | Spanda is not |
|---|---|
| The coordination layer for autonomous systems | A replacement for Python, C++, or Rust |
| A safety-first AI orchestration language | A general-purpose application language |
| A deploy-time verification tool | Just another robotics framework |
| Open source and community-driven | A proprietary platform lock-in |
Spanda orchestrates existing ecosystems. Train models in Python. Write drivers in C++. Deploy coordination logic in Spanda.
The first public release establishes credibility:
This is the foundation. The vision scales from a single rover to fleets of autonomous systems — all speaking Spanda.
Spanda is open source under Apache-2.0. We welcome contributors, roboticists, safety engineers, and AI researchers.