Intercognitive Protocol
How will trillions of Phygital Agents coordinate without centralized control?
Smarter, Faster, Stronger — coordinating and improving collectively across the planet at the speed of light. Not someday. Physical AI infrastructure is hardening into open standards now.
The Intercognitive Foundation creates interoperable standards for physical AI, enabling coordination across decentralized systems. This is the Agent Protocol that serves as the physical instrument for robotics at scale.
Flow of Commerce
Intercognitive implements the same six-stage commerce flow as digital agent protocols: DISCOVER → COMMUNICATE → COMMERCE → AUTHORIZE → SETTLE → EVALUATE.
| Commerce Stage | Pillar Layers | Function |
|---|---|---|
| Discover | Identity (pillar 1) | Machine publishes self-sovereign credential via peaq DID |
| Communicate | Connectivity (pillar 7) + Sensors (pillar 4) | Fleet signals state, environment, availability |
| Commerce | Maps (pillar 3) + Positioning (pillar 5) | Verified location enables service pricing and routing |
| Authorize | Fees (pillar 2) | Verifiable Intent constraints applied — same eight types |
| Settle | Compute (pillar 6) + Standards (pillar 9) | Edge compute processes; interoperability rules govern settlement |
| Evaluate | Orchestration (pillar 8) | Swarm feedback — coordination quality feeds next cycle |
Universal Machine Time is the prerequisite for every stage. You cannot verify position without verified time.
See Where Tracks Converge for the full digital↔physical equivalence map.
Architecture
The Foundation is developing a comprehensive standard framework:
| Component | Function | Robotics Application |
|---|---|---|
| Identity | Self-sovereign passports for connected machines | Know which robot did what |
| Fees | Peer-to-peer transaction systems | Robots get paid for work |
| Maps | Decentralized navigation data | Know the terrain |
| Sensors | Standardized perception data | Interoperate with environment |
| Positioning | Location data (RTK precision) | Navigate precisely |
| Compute | Decentralized AI backbone | Process at edge and cloud |
| Connectivity | Network links | Communicate with fleet |
| Orchestration | Multi-robot coordination | Swarm behavior |
| Standards | Interoperability rules | Work with any system |
The DePIN model works: GEODNET built the world's largest RTK network in under two years using token incentives. Centralized approaches cannot match that speed. The question is who controls the standards when billions of physical agents need to coordinate.
Foundation Members
Four organizations with specialized expertise:
| Organization | Expertise | Contribution |
|---|---|---|
| Auki Network | Posemesh — decentralized machine perception | Spatial data exchange |
| GEODNET | World's largest RTK network (12,000+ stations) | 100x GPS precision |
| Mawari | Real-time 3D data for AR/VR/spatial computing | XR infrastructure |
| Peaq | Web3 Economy of Things on Polkadot — Purple Paper | Blockchain layer + UMT + machine identity |
Broader ecosystem: Helium, Hivemapper, NATIX Network
Pillar Interfaces
Each pillar is an interface — a defined capability other agents can rely on. Read this as a reference: when agent A interacts with agent B through pillar X, this is what X carries.
| Pillar | Provides | Interface |
|---|---|---|
| Identity | A cryptographically verifiable machine passport | Self-sovereign DID resolvable onchain; signed messages |
| Fees | Machine-to-machine value exchange without prior contract | Published price schedule; open-ledger settlement |
| Maps | A shared model of the physical world | Open map schema; decentralized delta contributions |
| Sensors | Perception data any agent can interpret | Standardized, versioned schema with provenance and integrity |
| Positioning | Location with a known error bound | Verifiable source (RTK, VIO, fused); declared uncertainty |
| Compute | Inference with traceable origin | Declared location (edge, local, remote) and model hash |
| Connectivity | Network presence across vendor boundaries | Open network standards; graceful offline degradation |
| Orchestration | Coordination between agents that never met before | Intent and state exposure; authorized coordination messages |
| Standards | Interoperability across versions and vendors | Versioned interfaces; rejection of malformed messages |
| Time (UMT) | A shared clock every action can be anchored to | Verified time source; onchain PTP sync |
Time is listed explicitly. Every other pillar depends on it — you cannot prove what or where without first proving when. The robotaxi scenario below fails at step 2 if either vehicle's clock is unverified.
Universal Machine Time (UMT)
Peaq's March 2025 contribution — the first onchain implementation of Precision Time Protocol (PTP). Any machine or DePIN can sync its clock to the network and timestamp events at nanosecond-grade precision.
- Onchain PTP — verified timestamps, nanosecond target
- Critical for autonomous vehicles, robotics, sensor networks
- Foundation every other pillar assumes
"You can't understand space without first understanding time. Careful time calibration is how positioning technologies like GPS work in the first place." — Mike Horton, GEODNET
Stack Integration
The standard integrates with:
- Peaq — Economy of Things — Machine identity, payments, and settlement
- Blockchains — Settlement and proof layer
- w3bstream — IoT data infrastructure
Robotaxi Example
A passenger books a ride across a city. Robotaxi A (fleet operator X) accepts, picks them up, drives 12 km, then hits a low-battery threshold near a zone boundary. Robotaxi B (fleet operator Y, different vendor) is 400 m away with capacity. The ride must hand off mid-journey — passenger, fare, liability, route — without either operator trusting the other in advance.
Walk the protocol:
| Step | Pillar(s) | What happens |
|---|---|---|
| 1 | Identity | A and B each resolve the other's DID onchain. Fleet ownership, insurance status, safety record are cryptographically verifiable — no vendor API required. |
| 2 | Time (UMT) | Both vehicles stamp their state against onchain PTP. The handoff window is defined in absolute time, not "A's clock said so." |
| 3 | Positioning | A publishes its position with an RTK source and uncertainty bound. B verifies it matches its own observation of A within the same reference frame. |
| 4 | Maps + Sensors | A and B share a common map schema and standardized perception data. B can interpret A's environment model without translation. |
| 5 | Orchestration | A proposes the handoff; B accepts. A third agent (or contract) witnesses the intent and the agreed meeting point. |
| 6 | Fees | The original fare is split by distance. Settlement is machine-to-machine against the published schedule. No bilateral contract between operators X and Y existed before this moment. |
| 7 | Connectivity + Standards | The entire exchange runs on open network and message standards. Either vehicle could be swapped for a third vendor's robot next week. |
What breaks without Intercognitive:
- No shared identity → X and Y need a bilateral integration deal before any handoff is possible
- No UMT → disputes over "when did the handoff actually occur" are unresolvable
- No shared map/sensor schema → B cannot safely inherit A's environment model
- No open settlement → fare split requires a clearing house or manual reconciliation
Without the protocol, the handoff requires months of business development per operator pair. With it, any two compliant robots can coordinate in real time the first time they meet.
Applications
The same pillar interaction applies across domains:
| Domain | Coordinating event | Pillars that carry the weight |
|---|---|---|
| Autonomous vehicles | Cross-fleet handoff, right-of-way negotiation | Identity, Time, Positioning, Fees |
| Smart cities | Shared infrastructure access (charging, curb) | Identity, Fees, Orchestration, Standards |
| Robotics fleets | Multi-vendor task allocation | Identity, Orchestration, Compute, Standards |
| Spatial computing | Cross-platform perception sharing | Sensors, Maps, Positioning, Time |
Context
- Agent Protocols — The commerce flow this protocol serves as the physical track
- Economy of Things — Machine identity and settlement layer
- Robotics Industry — Where this protocol applies
- DePIN Devices — Physical infrastructure incentivized by tokens
- Verifiable Intent — Same delegation chain applies to machine authorization
- Space Industry — Extension to orbital scale
- The What-Next Algorithm — The three loop types applied to physical AI: runaway (no shared standard), corrective (bilateral deals), virtuous (open protocol any robot inherits)
- Essential Algorithm — The nine pillars implement
INTENT(DID identity) →ROUTE(maps + positioning) →INFRASTRUCTURE(compute + connectivity) →SETTLE(fees + UMT) →FEEDBACK(orchestration) - Routes — The robotaxi handoff as Fork (which robot) + Obstacle (no shared identity without protocol) + Sign (UMT-verified settlement) + Bridge (open standard any future robot inherits)
Links
Questions
In the robotaxi handoff, which pillar is the single point of failure — the one that, if absent, makes every other pillar useless?
- If two fleets can resolve each other's identity and settle fares but share no map schema, does the handoff still work — or does perception mismatch make it unsafe?
- Universal Machine Time anchors every other pillar. Is that a dependency that makes the stack brittle, or the reason the stack holds together at all?
- The protocol assumes agents that have never met can coordinate in real time. What class of interaction still requires prior agreement between operators, and why?
- GEODNET built 12,000 RTK stations in two years with token incentives. Which of the remaining pillars has the weakest incentive design today, and what would fix it?