ServicesCase studiesAbout usLet's talk
Case study · Industrial robotics · IoT · Secure remote operations

The software layer inside a robot that manufacturers actually let onto their floor.

A multinational robotics manufacturer picked U-centrix as their main software partner and — under an NDA that keeps their name off this page — handed us the cloud and IoT layer, the operator platform, and the mobile experience their fleet runs on. From the browser an operator uses to take over a robot on another continent, to the digital face a floor worker reads at a glance, we built every layer of software behind the machine.

Fleet OSConfidential · robotics manufacturer
Under NDA
Sector
Industrial robotics · Manufacturing
Engagement
Main software partner
Cameras / robot
6
Recorded frames
0
Stack
Cloud & IoT layer · web + mobile · RBAC · multi-layer auth · live video
6
live camera feeds streamed per robot, straight to the operator
0
video frames recorded or stored — by design, not policy
2-way
operators guide the fleet, floor workers talk back to it
1 face
shared visual language for every robot on the floor
FLEET-OS · CONTROL ROOM · OPERATOR: J. KOVAČ● SECURE LINKROBOT-04 · LIVE CAMERASCAM 01 · FRONTCAM 02 · REARCAM 03 · LEFTCAM 04 · RIGHTCAM 05 · TOPCAM 06 · GRIPDIRECT STREAM · NO MEDIA SERVER · NOTHING WRITTEN TO DISKROBOT-04 · STATUSAnalyzing environmentobstacle detected · re-routingZONEAssembly Bay · Line 3BATTERY82%FLEET · 12 UNITSRobot-04NEEDS INPUTRobot-01RUNNINGRobot-02RUNNINGRobot-03RUNNINGRobot-05RUNNINGRobot-06CHARGINGOPERATOR ACTIONTake remote control →

One operating system for a whole robot fleet.

We started with discovery workshops to map the technical requirements, the target market, and every stakeholder involved in deploying, operating, and supporting an autonomous robot on a manufacturing floor.

What came out of that was a robot fleet operating system: secure remote-control capabilities paired with an interface that lets workers and robots communicate directly on the floor. The remote-control layer exists mainly as a safety net — a way for a human to step in the moment a robot needs help.

Through it, operators configure and deploy robots, and — when something goes wrong — take control remotely, resolve the issue, and hand the robot back its task. The same system streams operational data, video, and sensor feeds live to an operator sitting in a central control facility, often nowhere near the plant floor the robot is actually working on.

We owned this end to end: the cloud and IoT layer underneath, the web platform operators use, the mobile experience, the networking and system architecture, and the integration between the browser-based interface and the robot's own operating system. From that browser, an operator can guide, monitor, trigger actions on, and — in defined scenarios — take full control of a robot on the other side of the world.

Everything had to hold up in someone else's factory.

Manufacturing companies protect their operations fiercely. Filming, recording, or copying data on-site is usually locked down hard — any leak could hand a competitor real strategic information.
Constraint 01

Facility-grade confidentiality

Recording, copying, or photographing anything inside the plant was off the table by default. Every design decision had to assume the strictest possible restriction, not the most convenient one.

Constraint 02

Latency a human can trust

An operator on the other side of the internet still needs to steer accurately and safely. Even a small delay between what the robot sees and what the operator sees makes remote control genuinely dangerous.

Constraint 03

A robot that can't be hijacked

Role-based access, multiple authentication layers, and hardened networking were non-negotiable — a manipulated robot on a live floor is a safety incident, not a bug report.

Constraint 04

Workers who feel safe near it

A robot people can't read is a robot people don't trust. The interface had to make its intentions legible to anyone standing near it — no manual, no training session required.

Video that's never recorded — by design.

Rather than harden a media server against misuse, we removed the thing that could be misused: the video never passes through one, and it never touches storage.
All six on-robot cameras stream directly to the authorised operator, in real time, with nothing recorded along the way
ROBOT-046 CAMERASCAM 1CAM 2CAM 3CAM 4CAM 5CAM 6ENCRYPTED · REAL-TIME · P2PNO MEDIA SERVERNO DISK · NO CACHEAUTHORISED OPERATOR6-feed live viewrole-checked · MFA · RBACACCESS LAYERSIdentity checkRole & scopeSession auth✓ Stream granted

Give the robot a face, and people stop guessing.

We designed a digital face, displayed on the robot itself, that shows its current state and its next move — so a worker walking past knows what to expect without reading a manual.
Idle
Awake, waiting for its next task
Analyzing
Reading the environment before acting
Blocked
Needs a person to clear the way
Moving right
Direction shown before it moves
Needs help
Stuck on an obstacle, calling an operator

Operators guide it. Workers talk back.

The finished system runs in both directions — a control room can take over a robot from anywhere, and someone standing next to it on the floor can redirect it without touching a computer.
Control room

Operator console

LIVE · CAM 01SENSOR FEED4.2m/s²ACCESS LEVELSenior operatorRobot-04 requesting assistanceAssembly Bay · Line 3Take over
  • Configure and deploy robots across the fleet
  • Watch six live feeds and sensor data per unit
  • Take remote control the moment something needs a human
Shop floor

Worker interface

Robot-04 is blockedClear the pathReassign taskGive it more information →
  • Read the robot's state and next move at a glance
  • Feed it context when it hits something unexpected
  • Reassign or unstick a task without calling the control room
— Let's talk —

Building something that has to be this secure?

A 30-minute call with our CEO or technical lead. No sales script, no obligation — just a real conversation about what you're trying to solve.

Book a call