# Robotics — Otonoma

Otonoma's network puts robots, drones, people and software from different vendors on one plan — designed in simulation, overlaid on the systems you already run, supervised by people.

Robotics has long suffered from vendor siloing. Otonoma's network orchestrates robots, drones, people and software from different vendors as one fleet — designed and tested in simulation first, laid over the systems you already run, with people supervising the decisions that matter.

[Request a demo](https://www.otonoma.com/demo-request).

## Videos: see it run — in simulation first

These are simulations and demos, many built with NVIDIA Omniverse (Otonoma is a registered NVIDIA development partner). They show how the network, working with agentic artificial intelligence (AI), plans and coordinates mixed fleets before anything touches a real floor. Channel: [youtube.com/@theotonoma](https://www.youtube.com/@theotonoma).

- [Autonomous tracking, swarming and routing](https://youtu.be/q1VeeUnXRQg) (simulation, April 2026) — a simulated fleet tracks, swarms and routes as one team. Each vehicle is an actor on the network, and a shared plan coordinates them, not one pilot per machine.
- [Robot "partners" at a conference we couldn't attend](https://youtu.be/b1f4PNnnrNw) (demo, March 2026) — Otonoma couldn't be at NVIDIA's developer conference in person, so it orchestrated robotic stand-ins to attend virtually. An agentic runtime calls the network to coordinate different robot types and drones together.
- [Agentic scenario running with NVIDIA Omniverse](https://youtu.be/9JRHq3qnM5I) (simulation, March 2026) — the network runs scenarios inside a digital twin, so a plan is exercised end to end before it reaches real equipment.
- [Agentic AI drives scenario creation](https://youtu.be/-baGghdwtfo) (simulation, February 2026) — describe the situation and agentic AI builds the scenario in NVIDIA Omniverse: the space, the machines and the tasks, ready to run.
- [50 robots with look-ahead planning](https://youtu.be/mD8uKfm6agQ) (simulation, February 2026) — fifty simulated robots plan ahead to stay out of each other's way. How many robots does a facility really need? Why guess — run the scenario.
- [Dynamic goal trees: drone fire suppression](https://youtu.be/T_5VrTMRd94) (simulation, October 2025) — as new fires appear, the goal tree changes and the drone team re-plans on the fly, splitting up to tackle several blazes at once.
- [Simulating human response to a robot malfunction](https://youtu.be/0gWFJlbRwVk) (simulation, August 2025) — a robot fails mid-task in NVIDIA Omniverse, and the simulation includes how the people nearby respond, because people are actors on the plan too.
- [Bridging information technology (IT) and operational technology (OT)](https://youtu.be/gavv_h3QtgY) (simulation, July 2025) — warehouse robots in NVIDIA Isaac Sim take their work from business systems through the network, closing the IT–OT gap on the warehouse floor.

## The problem: robots from different vendors don't work as one team

Each fleet arrives with its own controller, its own software and its own map of the world. Getting them to cooperate — with each other, with the people on the floor, and with the business systems that decide what should happen — is still bespoke work, facility by facility.

- **Vendor siloing.** Robots, drones and automated equipment from different manufacturers can't share a plan, so every mixed fleet becomes a custom integration.
- **Multi-use and humanoid robots.** As robots take on many jobs instead of one, the hard part moves from controlling a machine to coordinating many machines across changing work.
- **Coordination at scale, amid change.** Plans break when a robot fails, an order changes or a person steps in. The fleet has to re-plan together, safely, without a rewrite.

## How it works: intent, to tasks across many machines, to human supervision

You say what needs to happen. The network turns it into tasks for the right machines and people, runs them, and keeps a person in charge of the calls that need one.

1. **Design in simulation.** Build and test the workflow in a digital twin, such as NVIDIA Omniverse, before it touches the floor. Try the edge cases there.
2. **Plan.** Intent becomes a plan across many machines. The plan is generated before execution, so it can be seen and validated up front.
3. **Operate.** Robots, drones, people and software carry out their parts. When something changes, the plan adapts; the judgment calls go to a person.
4. **Learn.** Every run and every human decision is recorded, and feeds the next version of the workflow — back in simulation first.

## Under the hood: actors can be software, hardware or human

On the intelligent network, every participant is an actor with skills it offers to the others. A warehouse robot, a drone, a scheduling system and a floor supervisor are all actors on the same plan. That is what ends vendor siloing: the network doesn't care who made the machine, only what it can do.

- **Work is matched, not wired.** A request can go to a persona — a pool of actors able to do the job — and the network picks one by availability, cost or time.
- **You see the plan before it runs.** Plans are generated before execution, so they can be inspected and validated up front, then executed the same way every time.
- **It survives failures.** Each node keeps persistent state and its own ledger, so work carries on when a node goes offline.
- **Humans in the loop, by design.** People are actors too. Approvals, overrides and hand-offs are part of the plan, not an afterthought.

## Brownfield first: overlay the robots and systems you already run

No rip-and-replace. The network sits on top of existing infrastructure — the robots, the people and the legacy software — and coordinates across them.

- **Multi-vendor fleets.** Robots from different manufacturers work to one plan.
- **Simulation-ready.** Workflows are tested in a digital twin with NVIDIA Omniverse and Isaac Sim before deployment.
- **Built for the edge.** The node is written in Rust, has a small footprint and runs on constrained hardware. After a reset it resumes where it left off.

## Fleets and swarms: from one facility to the sky above it

- **Facilities.** Warehouses and factories where fixed automation, mobile robots, humanoids and people share a floor and a plan.
- **Drones and swarms.** Tracking, routing and swarm decisions coordinated in real time — moving beyond one pilot per vehicle.
- **Response.** Missions like fire suppression, where the goal changes minute by minute and the team has to re-plan together.

## Robots are part of the trip, too

Hotels, airports and venues already run service robots for delivery, cleaning and baggage. On their own they are one more system that doesn't talk to the others. On the network, a robot is simply another actor — on the same plan as the flight, the car and the front desk. So when a guest's flight is 90 minutes late, the towels and the late supper arrive when she does, not when the original booking said.

- **Hotels.** Room deliveries and housekeeping robots timed to the real arrival.
- **Airports.** Baggage and cleaning fleets that re-plan with the departures board.
- **Venues.** In-seat delivery that follows the fan to a new seat.

See the hotel side: https://www.otonoma.com/travel/hotels

## Travel is first. It is not the market.

The network that re-plans a trip when a flight is late is the same one that coordinates a mixed fleet on a factory floor. Travel is where Otonoma is building actors first; the platform underneath is the same everywhere.

- [Other use cases](https://www.otonoma.com/use-cases)
- [Build on the network](https://www.otonoma.com/developers)
- [Travel & Hospitality](https://www.otonoma.com/travel)

## Put your fleet on one plan

Test it in simulation before it touches the floor. Otonoma is working with a small number of partners to build it. Several prototypes are in operation today.

- [Request a demo](https://www.otonoma.com/demo-request)
- [Contact](https://www.otonoma.com/contact)