How to use Codex to build a robot mission simulation
Codex built a live robot mission simulation with checked routes, memory, and audit evidence; its owner kept optional physical camera access strictly read-only.
devpost.com
Suggested prompt
Suggested prompt written by Agents In Use, not the source author. Replace bracketed inputs and review tool permissions first.
Extend this robot mission simulator with a shared metric map, server-authoritative jobs, sequenced WebSocket poses, and collision-checked routes. Preserve inspectable memory and audit evidence. Keep the physical camera adapter read-only, label simulated telemetry, and verify stop/restart, wall clearance, browser behavior, and the Docker setup before my review.
Reach people exploring AI agents
See placements →Evidence at a glance
Sourced account- Source author
- Varianyk Danil
- Original account
- Read the source ↗
- Published
- Source link checked
- Reported result
- NEXUS ran a live three-robot mission simulation with checked routes and reported backend, frontend, navigation, and browser verification; no physical motor authority was enabled.
- Human involvement
- The human supplied goals, chose the read-only physical boundary, reviewed visible behavior, and retained all real-robot decisions.
A linked account supports attribution. It does not mean we independently reproduced its result.
Report an issue with this case
Open correction emailOpens your email app with the case link included. Nothing is sent automatically.
How the agent did it
- Codex migrated the repository into a repeatable macOS and Docker development setup.
- It implemented mission state, memory, live virtual-robot telemetry, route checks, and the browser console.
- The owner reviewed browser behavior and verification while keeping the optional physical adapter read-only.
Varianyk Danil used Codex to turn an existing Windows-oriented repository into NEXUS, a repeatable robot-mission simulation on macOS and Docker. The work joined a FastAPI control plane, PostgreSQL memory and audit data, a browser console, and a C++ navigation component.
Three virtual robots moved through a shared metric building model. Codex helped implement sequenced WebSocket telemetry, route checks, mission stages, and the interface that showed why a robot was assigned to a job. The team exercised the final browser experience and checked route geometry so movement did not simply cut through walls.
The owner selected a clear physical boundary. A Spot integration could inspect identity, battery, and cameras, but could not acquire motor authority. The deterministic demo worked without a model API key, and simulated telemetry was not presented as a measurement from a real fleet.
The reported result was a running, tested simulation with visible jobs, memory, paths, and audit evidence. That is distinct from a deployed robot controller. Physical pilots were still future work, dependent on further authentication, grants, watchdogs, and safety review. The account shows Codex implementing an inspectable simulation while the human retained product and physical-operation decisions.
Prompt
Suggested prompt (written by Agents In Use, not the author's)
Source: Devpost · Varianyk Danil
Frequently asked questions
Was this a real robot fleet demonstration?
The moving fleet was virtual; physical telemetry was kept separate.
Could the Spot adapter command motion?
No. It did not acquire a lease, power motors, or send movement commands.
Did the deterministic demo need an API key?
No. The optional model review was separate from the complete demo.
More details
About this account
Codex migrated the repository into a repeatable macOS and Docker development setup. It implemented mission state, memory, live virtual-robot telemetry, route checks, and the browser console.
Original developer account opened; linked repository could not be fetched. This is explicitly a simulation-first foundation; no certified robot controller or physical motor demonstration is claimed. Human review and API-model review have different roles. Suggested prompt written by Agents In Use; no original task prompt was published in the verified account.
Credit: Varianyk Danil — Devpost
Source published
Added to Agents In Use
Source checked
Tools and topics
Codex · FastAPI · PostgreSQL · Three.js · Docker · C++
digital twin · live telemetry · simulation boundary
Editorial completeness: 100/100. Measures documented fields, not agent performance or verified outcomes.