Labs: pilots and prototypes on a test site, the next generation of the network tried first here. The loop, step by step: 1. A tower, a radar and robots on a range. 2. A new idea goes onto the range. 3. The radar and tower collect the first data. 4. Each reading is stamped with time and place. 5. Data streams to the lab compute. 6. Relays carry the packets without reading them. 7. The drone flies the test pattern. 8. Instruments confirm the response took effect. 9. Signatures are checked before any model runs. 10. Results are compared with the baseline. 11. The edge releases confirmed results only. 12. The lab publishes the finding. 13. The alert arrives with its evidence attached. 14. The whole chain is archived for audit. 15. Everything settles back to idle.

01 / 15

A tower, a radar and robots on a range

Nexus Observatory

/

Research pilots that put new instruments, radios and models on real sites, open benchmarks that compare them on the same data, and a path from a paper result to a qualified release, with the method published either way.

At a glance

The frontier, taken into the field

Frontier labs runs research pilots on real sites, open benchmarks that compare methods on the same data, and a path from a paper result to a qualified release, with the method published either way. Every pilot leaves evidence, whatever the outcome.

The problem

Results that never leave the paper

New sensing, radio and AI methods are published every month. Few are tested on a real site, fewer still become something an operator can buy and trust. The distance between a promising result and a dependable service is crossed by testing, not by announcements.

Components

• Pole-mounted edge gateway node
• Compact X-band weather radar / ISAC sensing node on a lattice tower top
• Atmospheric lidar station
• Particulate monitoring station and sensor node
• Private-5G site
• "AI tower in a box"
• Data hall row pair with cold-aisle containment
• Operations console with a video wall

Interfaces

Instruments

SDI-12 · RS-485 / Modbus RTU · 4–20 mA

Radio

Ethernet · LTE-M / NB-IoT · LoRaWAN

Time and position

GNSS 1PPS

Security

TPM 2.0

Publishing

MQTT

Where it works

• New technology, a real site, a written plan
• Sensing and communication on shared compute
• Frontier labs: where it should stand
• Frontier labs: what it can reach

In the story below

1. The system, part by part
2. Assembly
3. Installation
4. Deployment
5. Connections
6. Operation
7. Failure and recovery
8. Scale and federation
9. What you get

Scenes · 16

Next steps

1. The problem

   Results that never leave the paper

   New sensing, radio and AI methods are published every month. Few are tested on a real site, fewer still become something an operator can buy and trust. The distance between a promising result and a dependable service is crossed by testing, not by announcements.

2. What it does

   The frontier, taken into the field

   Frontier labs runs research pilots on real sites, open benchmarks that compare methods on the same data, and a path from a paper result to a qualified release, with the method published either way. Every pilot leaves evidence, whatever the outcome.

3. Pilots

   New technology, a real site, a written plan

   Each pilot has a written plan, reference measurements and success criteria before it starts, and its report is published whatever the outcome. Plans state the hypothesis, the method and the data that will decide the outcome, so nobody can move the goalposts afterwards.

   See Pilots

4. AI-RAN and 6G

   Sensing and communication on shared compute

   Radios that also sense, and networks that run AI beside the RAN, are tested on live sites with real traffic. Results are measured against the same site's normal traffic, so gains are real and not artefacts of a quiet day.

   See AI-RAN and 6G

5. Design it in the Field Deployment Studio

   Frontier labs: where it should stand

   A deployment of frontier labs starts from the ground: candidate spots are staked and the viewshed from each mast top shows what is in sight over the terrain. The spot that sees the most of what matters, with access and power within reach, becomes the first site of the frontier labs design.

6. Design it in the Field Deployment Studio

   Frontier labs: what it can reach

   From that spot, the coverage of the frontier labs radio is computed over the real ground, with its isolines and the shadow behind every ridge. Where the frontier labs coverage leaves a gap, the design adds a site before anything is bought.

7. Design it in the Field Deployment Studio

   Frontier labs: will the link close

   The hop from a frontier labs site to the hub is drawn on its terrain profile, with the first Fresnel zone to scale and the worst clearance marked. The link budget of the frontier labs design closes from transmitter to margin with its assumptions stated, or the design moves the mast.

8. Design it in the Field Deployment Studio

   Frontier labs: will it last the winter

   The solar array and battery of each frontier labs site are sized from the yield of the darkest month against the load. The state of charge of the frontier labs design never crosses its floor, and the whole plan goes to the team as a manifest of sites, links, power and parts.

   Open the Field Deployment Studio

9. Space and NTN

   Direct-to-device and new payloads, tested

   New non-terrestrial links and payloads are trialled against ground truth before anyone builds a service on them. Links are measured against independent references, and the measurements are published with the method used. Ground stations, terminals and reference receivers record every pass for later analysis.

   See Space and NTN

10. Into the product

    From a result to a qualified release

    A pilot that succeeds goes through qualification and joins the product as a release, rolled out like any other. The release keeps the pilot's evidence attached, so operators know what was shown and where it was shown. Nothing reaches the fleet unqualified.

    See Into the product

11. Safe experiments

    Pilots that cannot harm the fleet

    Pilot equipment joins in its own segment with its own scope, attested like any other node, so an experimental radio, model or payload can never disturb the production network. When a pilot ends, its access is revoked and its data is kept or released as the plan states.

12. Benchmarks

    Open data, one method, comparable results

    Benchmarks fix the data and the method so results from different teams can be compared honestly. Datasets, scoring code and baselines are published, so anyone can reproduce a score or challenge it. Scores are reported with their uncertainty and the conditions under which they were measured.

    See Benchmarks

13. Physical AI

    Drones and robots, inside published red lines

    Autonomous machines are tested under published limits and supervision, with every mission logged. Limits, supervision and incident reporting are written before the first flight or run, and reviewed after it. Every mission log is kept, and any breach of a limit stops the programme until it has been reviewed and the limit reassessed.

    See Physical AI

14. How it scales

    From one pilot to the fleet

    What works on one site is repeated on a few, then released to the fleet in rings, with the pilot's evidence attached. Wider rollouts are gated on the same health checks as any other release, and can be rolled back the same way.

15. What you get

    Pilots and prototypes, under contract

    Benchmark methods and pilot reports, published either way, are open. Research pilots and prototype engineering are contracted; pilots are scoped and priced individually, and their results are shared with the partners involved. Prototype engineering turns a successful pilot into a design ready for qualification and supply.

    Partner with usTalk to sales

16. Next step

    Propose a pilot

    Bring us the method, the question and the site. We write the pilot plan together; the pilot plan states the question, the site, the measurements and the criteria for success before any work begins. Partners keep their own rights in the results, as agreed before the pilot starts.

    Plan a deploymentPartner with us

01 · What this platform covers

Frontier labs, area by area.

Choose an area for its systems, standards, the industries it serves and how we deliver it.

02 · Built on the open protocol

One protocol under every platform.

Marked: the functions this platform leans on hardest. The protocol itself is open, so what you build here runs anywhere the protocol runs.

01 JOIN

02 ADMIT

03 DESCRIBE

04 TIME

05 OBSERVE

06 QUALIFY

07 FEDERATE

08 ORCHESTRATE

03 · Open and enterprise

Open by default. Enterprise by contract.

The open layer lives on the main site, free to use and build on. Enterprise services are delivered under contract by Nexus Observatory Inc.

Open, on the main site

• Benchmark methods
• Pilot reports, published either way

Enterprise, on this site

• Research pilots
• Prototype engineering

04 · Industries served

The industries that build on frontier labs.

Each industry has its own site: its use cases, its community and the enterprise path, with this platform among the systems behind it.

05 · Every platform

Every platform, one architecture.

One platform for each technical area of our work, on the same protocol, the same data and the same design.

Bring us the site.

We bring the survey, the design and the crew. Nexus Observatory Inc. The form names who receives your details.

More presentations

See it from every side.

Each opens as a full-screen presentation: deeper pages on the same subject, explainers, stories, guided tours of the places, network scenes and the observatories that came before.

In depth

Explainers