Skip to main content

Your robots, ready for factories, ports, manufacturing plants and mines.

Measure reliability. Find the root cause. Before every rollout, policy, deployment and change.

Get started free

Free for macOS, Linux and Windows. Need a pilot? Book a 30-minute call ↗

RUNFP / ONE-MINUTE OVERVIEW

In the field with

  • PILOT_01Warehouse robot fleet
  • PILOT_02Actuator manufacturer
  • PILOT_03Robot AI lab
  • PILOT_04Systems integrator
  • PILOT_05Drivetrain maker
  • PILOT_06Industrial QA line
  • PILOT_07Mobile robot OEM
  • PILOT_08Manipulation startup

01 / RUNFP · THE COMPLETE RUN HISTORY

Every run.
All the evidence.

RunFP plugs into the robots you already run and stays on your premises. It ties every result to its code, model, hardware and setup, then measures reliability across all your runs: evaluate each new policy against the last, spot regressions, replay every failure.

CONTEXT FROM EVERY RUNLOCAL WORKSPACE / ILLUSTRATIVE DATA

Swipe to inspect all 7 sources →

OORB / RUNFP480 RUNS · 2 LIDAR STACKSREGRESSION FLAGGED

THE EVIDENCE LAYER

480 runs
RUN
ARCHIVE
BaselineSaved runs
New LiDAR stackSaved runs
EXAMPLE EPISODERecorded + judgedSuccess Failure
RUNTIME DATA / source.mcap × 480

Compare signals across hundreds of saved runs and return to any failure for replay.

Run archive
480 saved recordings
Comparison
baseline → new LiDAR stack
Finding
perception confidence −4%
480 runs capturedBaseline comparedPerception change flagged

RELIABILITY SCORE

EXAMPLE 01 / 08

Pick + return

92.5%

Task success
across 480 runs

Baseline95%
New LiDAR stack90%

ROOT CAUSE

SAME EXAMPLE

Perception confidence dropped

−4%perception confidence

Perception confidence dropped by 4% with the new LiDAR stack.

REPLAY / SAVE / LEARN

Replay and save every failure to learn from it.

One run, replayed

Every run leaves a trace.

A recorded pick-and-return, played back. RunFP follows the robot’s own signals and checks each step of the task as it happens.

RUN 0412 / PICK + RETURN00.0 s / 12.0 s
  • Acquire
  • Lift ≥ 20 cm
  • Hold ≥ 1 s
  • Return + release
Phase
Approach
Lift
0.0 cm
Grip
80 mm

RUN 0412 · Waiting for the acquisition step.

02 / THE DEPLOYMENT GAP

From demo
to deployment.

Robots running AI never behave the same way twice. When a deployment slips, engineering, the integrator and the site each have a theory, and nobody can show whether a change made things better or worse. How to close that gap is in our guides to robot deployment and robot policy evaluation.

  1. 01

    How reliable is this deployment?

    A confidence score per task and per operating condition, across every run, with the conditions nobody has tested listed as untested.

  2. 02

    What changed when it failed?

    The root cause, ranked: code, model, dependencies, hardware, environment or setup.

  3. 03

    Can your customer trust the numbers?

    A record of every run, kept on your premises, that your team and your customer can both check.

Reliability vs. engineering effort
Reliability at each step of engineering effort
Effort stepReliability
12%
28%
318%
432%
548%
664%
776%
884%
989%
1090%
1191%
1292%
1393%
1495%
1596.5%
1698%
1799.2%

ILLUSTRATIVE CURVE / THE EFFORT BETWEEN A WORKING DEMO AND A DEPENDABLE DEPLOYMENT

FROM A PILOT / END-OF-LINE QA

4 days → 20 minutes to root cause
30% of an actuator maker’s units started failing on the QA line and stayed that way for weeks. Engineering, the integrator and the site each had a theory. RunFP traced it to an HVAC unit reporting the wrong temperature.

Mobile Station

Eyes that move with your fleet.

A fleet supervisor on four wheels that drives with your robots. Its mast rises to see over the work, and it records the conditions of every run and syncs them with RunFP, wherever the fleet goes.

⭐ Must Read: News Update - ประเด็นร้อน⭐ Must Read: News Update - ประเด็นร้อน01360° camera⭐ Must Read: News Update - ประเด็นร้อน02Depth camera⭐ Must Read: News Update - ประเด็นร้อน032D lidar⭐ Must Read: News Update - ประเด็นร้อน04Ambient light⭐ Must Read: News Update - ประเด็นร้อน05IMU⭐ Must Read: News Update - ประเด็นร้อน06Edge compute
FWD0901802701M2M3MAMBIENT360°87°ON-PREMXYT012.2M 103T022.1M 203T032.2M 319T041.1M 141
Coverage, top-down1 ring = 1 m

Simulated // readout

Mobile Station06 PARTS

What the Mobile Station carries

  • 5 sensors
  • 1 edge computer
  • Processed on board

Tracks00/04

04 / USE CASES

One reliability layer,
every kind of robot.

OORB / RUNFPUSE CASE 01 / 03

AMRs & mobile fleets

Mobile fleets in changing environments.

Live AMR experiment: from ROS bag to run breakdown.

OORB Mobile Station 3D blueprint

What breaks

  • Layout changes cause localization drift
  • Sensor topics drop to 0 Hz
  • Uneven motor wear across the fleet

What OORB records

  • ROS bags
  • Motor telemetry
  • TF / odometry
  • Topic rates
  • Firmware / config

One investigation, ranked by signal

RunFP ranked 7 changes against the baseline run. The leading signal: camera topics had dropped to 0 Hz.

Seconds

from ROS bag to run breakdown

NEXT / ROBOT ARMS & LEARNED POLICIES
01 / 03

Same task, next run

When a run breaks, you see why.

Joint feedback drops out for six seconds. RunFP marks the gap and reports the run as unverified, instead of guessing whether it passed.

RUN 0413 / PICK + RETURN00.0 s / 12.0 s
  • Acquire
  • Lift ≥ 20 cm
  • Hold ≥ 1 s
  • Return + release
Phase
Approach
Lift
0.0 cm
Grip
80 mm

RUN 0413 · Waiting for the acquisition step.

05 / YOUR DATA

Your data stays with you.

  • Runs on your premises

    RunFP installs on your own machines. Recordings, results and reports stay on your network.

  • Starts from what you record

    ROS bags, MCAP files and test-rig exports. No changes to your robot to get started.

  • A record you can show

    Every result keeps the code, model, hardware and conditions behind it, so a claim can be checked.

QUESTIONS WE GET

What does OORB do, in one line?

RunFP records every robot run with the code, model, hardware and conditions behind it, measures reliability per task and condition across all of them, and ranks what changed when results drop. The Station adds an outside view of the conditions around your robots.

Where does my data go?

Nowhere you don’t choose. RunFP runs on your own machines and your recordings stay on your network.

Which robots and data do you work with?

Mobile robots and AMRs, robot arms running learned policies, bimanual setups and end-of-line test rigs. We read ROS bags, MCAP files and test-rig exports. If you record something else, tell us.

How does the free reliability audit work?

Pick one robot behavior and share the run data you already have. We come back with how reliable it is by operating condition, where it fails, and what changed when it did.

How is it priced?

Per robot, billed monthly, quarterly or yearly. The audit is free, and we quote once we know your fleet.

Does OORB replace our fleet management software?

No. A fleet manager dispatches and routes your robots. OORB works beside it: RunFP measures how reliably the work gets done, and what changed when it stops. What robot fleet management covers.

Do I need the Station?

No. RunFP works on its own from the data your robots already record. The Station is our hardware: a small mobile unit that records the conditions around a run and syncs them with RunFP.

06 / LET’S TALK

Your next deployment.
Let’s make it work.

Start with a free reliability audit: share run data from one robot behavior and we’ll show you where it fails and why. Built for teams that make robots and the integrators who deploy them.

OR SEND US A MESSAGE

Prefer email? contact@oorb.io