Flagship Project · Machine Economy Lab, HTW Berlin

RoboPay

Payment infrastructure for AI agents in the physical world.

A trustless payment layer for autonomous machines — robots, drones, delivery vehicles — that confirms real-world task completion and releases payment from a Solana escrow smart contract, with no human in the loop.

~8s
Settlement Latency
GPS-triggered Solana escrow
Live
PoC Transactions
Real Devnet — May 2026
Open
Research Paper
Architecture · Security · Ethics
2026–29
HTW Berlin Project
Registered research project
About

What We Build

Industrial robotics is evolving from centrally controlled single machines into semi-autonomous fleets that accept tasks, consume resources, and deliver services — but settlement still needs a human to confirm the work actually happened. RoboPay closes that gap: machine-provable proof that a specific device performed a specific service, on untampered measurements, triggering payment with no human in the loop.

Proof of delivery is checked one of three ways — GPS geofence, QR handover scan, or an AI agent visually judging the delivery (delivery_agent.py) — each benchmarked against the others. An independent second-opinion service (x402_verifier.py) demonstrates the x402 machine-payment protocol as a fourth comparison point. Every path ends the same way: payment released from a Solana escrow smart contract, over a secure Staex MCC tunnel with zero public IP exposure.

Machine Arrives
Drone · PiCar-X · Letterbox
↓
Proof of Delivery
GPS geofence · QR scan · or AI-agent vision judgment
↓
Staex MCC Signal
Secure M2M tunnel · zero public IP
↓
SOL Released
Solana Anchor escrow · ~8s · benchmarked against x402 Protocol
Research Question

“How can AI agents and blockchain-based payment protocols enable trustless, autonomous economic transactions between machines and humans — and what are the technical, ethical, and societal implications?”

Work Program

Four Open Questions

F1Evidence

How can an edge-AI model's output be attested so that model version, executing device, and result can later be conclusively linked?

F2Partition Tolerance

How can proof and settlement continue locally during network loss and reconcile consistently on reconnect?

F3Economic Agency

Which governance and authorization models let a machine trigger transactions under delegated budgets — including revocation and liability?

F4Cost of Trust

What compute, energy, and latency does continuous attestation impose on constrained hardware — and where is the limit of practicality?

The contribution is not a single cryptographic primitive but the closed chain from sensor to settlement — tested under real conditions rather than in the lab. Unlike overlay networks (WireGuard, Tailscale, Nebula) that secure the transport, RoboPay secures the origin and integrity of the claim itself.

People

The Team

Prof. Dr. Alexandra Mikityuk
Prof. Dr. Alexandra Mikityuk
Supervisor
Professor HTW Berlin · CEO, Staex GmbH
Kevin Ehrentraut
Kevin Ehrentraut
Founder & Technical Lead
Customer Success · IoT Hardware · Staex GmbH
Xinyan Liu
Xinyan Liu
Hardware & Mechatronics
Industrial Engineering · HTW Berlin · Internship at ABEX
Yashdeep Singh
Yashdeep Singh
Business & Market Research
Digital Business · BHT Berlin · Business Development at Hydro Berlin
Research Scope

What We Investigate

01
Payment Protocol Comparison
GPS, QR, and AI-agent proof of delivery vs. the x402 machine-payment protocol — four autonomous trigger models benchmarked for delivery use cases.
02
Hardware Prototype
Physical build: autonomous vehicle (PiCar-X) and drone — both running RoboPay on Solana Devnet.
03
Ethics & Societal Impact
Legal implications, liability frameworks, and societal effects of fully autonomous machine-to-machine economies.
Progress

Milestones

May 2026
Complete
Indoor Proof of Concept
GPS-triggered Solana escrow with real Devnet transactions. ~8s settlement latency. Payment releases automatically when the machine reaches the delivery geofence — no human click.
→ Indoor PoC Video
Oct 2026
In Progress
M1 — Outdoor GPS Payment Demo (Unit C)
PiCar-X drives to a GPS coordinate. Haversine geofence triggers Solana Anchor escrow release automatically — no human click. RoboPay stack benchmarked against x402 Protocol as autonomous payment alternative.
Oct / Nov 2026
Planned
M2 — Letterbox Node (Unit A)
Raspberry Pi receives arrival signal via Staex MCC tunnel. Display shows: "Car arrived — SOL from Escrow released." Machine-to-machine, zero human steps.
Nov / Dec 2026
Planned
M3 — Drone with RoboPay (Unit B)
Drone carries identical RoboPay stack (RPi + GPS + Anchor client). Same Solana escrow logic — proves hardware-agnosticism. RoboPay compared with x402 Protocol as autonomous payment alternative. Outdoor flight test planned.
Hackathon & Conference

Where We're Submitting

Read the Full Research Paper

Architecture, prototype, payment mechanisms, security analysis, legal & societal impact.