- Home
- Applications
Applications · Six jobs, six pages, one starting point
Motion platform applications, sorted by what has to be convinced.
The same six-strut mechanism can shake a component, feed an experiment, tilt a cockpit or hold an instrument still on a moving deck. What changes between those jobs is not the axis count. It is what has to be convinced — an instrument, a method, a person or a sensor — and that decides the specification, the interface and what counts as proof.
Project sitting between two of them? That is the normal case rather than the awkward one — a test rig that has to feel right to an operator, or a simulator that also has to reproduce a recorded profile. Send the job and the first reply names which page owns it.
6DOF motion platform applications including industrial testing, research, flight simulation, driving simulation and motion compensation
The question that sorts this column
Five of these produce motion. One removes it. All six differ in what they have to prove.
Buyers usually arrive with an industry label — aerospace, automotive, marine, laboratory. It is the wrong first cut, because two projects in the same industry can need opposite machines while two in different industries need the same one.
The cut that works is what the motion has to satisfy. A test article does not care whether the motion feels right; it cares whether the commanded trajectory arrived at the measurement point. A person cares about almost nothing else. An experiment cares whether the same input can be produced again in six months. A sensor on a moving deck cares only about how much motion is left.
Those four are different acceptance tests, and an acceptance test written for the wrong one is where application projects fail. So each of the six pages below states three things in its own words: when to start there, what it asks for first, and what counts as proof.
Start from the physical outcome. The industry label comes second, and the axis count comes last.
Already know the outcome you need? Take it to the crosswalk below and pick the page →
Motion platform application proof criteria for industrial testing, research, simulators and motion compensation
The routing table for this whole column
The 6-Page Application Crosswalk
Read the second column until one row describes your project, then open that page. Each of the six sections further down expands its row and links to it.
| Application page | Start here when | What it asks for first | What counts as proof |
|---|---|---|---|
| Industrial Testing | A controlled trajectory has to be applied to a component, a fixture or a complete article, and something has to be measured while it happens | The article and its fixture as one moving assembly; the trajectory or the standard behind it; the measurement point | The commanded trajectory reproduced at the measurement point, logged and repeatable |
| Research Motion Platform | An experiment or a research programme needs a repeatable dynamic input, and the payload will change over the platform’s life | The load band rather than one payload; the host environment; the measurement plan | Run-to-run repeatability, and the same input reproducible years later |
| Motion Platforms for Simulators | A simulator that somebody else owns needs motion underneath it, and the platform is a supplied subsystem | The complete moving assembly including occupants; the cue set; host interface and timing | Platform acceptance at load, plus cue behaviour agreed in writing before it is built |
| Flight Simulator Motion Platform | A flight training device needs the motion system beneath a defined moving cockpit | Cockpit mass properties; the cue set; interface timing; the safety chain | Platform acceptance, plus whichever objective motion tests the programme is being evaluated against |
| Driving Simulator Motion Platform | A driver-in-the-loop or vehicle-research programme needs motion beneath a buck the customer owns | Buck mass properties; the driving task; the vehicle model or hardware in the loop; cue timing | Cue timing behaviour and platform acceptance at load, against criteria set before the build |
| Motion Compensation Platform | A measured or commanded base disturbance has to be reduced at a defined payload point | The base motion itself; the reference sensor and its latency; the residual target; saturation and fail-safe behaviour | Residual motion measured at the payload by an instrument that is not the platform’s own encoders |
How the six relate. Flight and driving are specialist children of Motion Platforms for Simulators — if the simulator is neither of those, or the cockpit type is not yet decided, start on the parent page. Industrial testing and research sit together because both use motion as an input to something being measured; they differ in whether the answer is a pass or a data set. Motion compensation stands apart because its control objective, its input signal and its acceptance metric all run in the opposite direction.
What none of the six sells. A complete simulator, a training device qualification, a test method, an accredited laboratory service, a vessel or a certified safety system. What is supplied on every one of them is the moving machine, its control, the interface contract and the acceptance run against it — the boundary is set out in the shared section below.
Two rows look like they both describe your project? Send the objective and one representative motion file, and the first reply names the page →
Application 01 · Motion as a test input
Industrial testing, where the motion is the stimulus and something else is the answer
A trajectory is applied to a component, a fixture or a complete article, and instrumentation records what the article did about it. The platform is the stimulus. Nobody is buying the motion for its own sake.
The distinction that page opens with is the one that saves projects: a motion platform is a position machine. It commands a pose and holds it against whatever the payload does. A shaker is a force or acceleration machine, single-axis and high-frequency, and it answers a different question. Asking a platform for an acceleration spectrum at a mounting point is a legitimate requirement, but it has to be written as an acceptance in the quantity you actually care about, because a platform that follows its commanded position perfectly can still miss an acceleration target once the fixture flexes.
The other thing that page insists on is that the fixture is part of the test article. Whatever holds your component to the table has mass, stiffness and a resonance of its own, and it sits between the machine and the measurement point. A test rig accepted on the table and used through a soft fixture is measuring the fixture.
The platform commands position. If your requirement is a force, say so before the acceptance is written.
Have a standard or a recorded profile to reproduce? Open the industrial testing page →
Application 02 · Motion as an experimental variable
Research platforms, specified for a decade of experiments nobody has designed yet
A laboratory buying a motion platform is not buying it for the current project. It is buying an instrument that has to still be usable when the students, the payload and the research question have all changed.
That changes what the specification is written against. A production test rig is sized for one article; a research platform is sized for a band — a range of payloads, a range of mounting heights, and an interface open enough that whoever inherits it can drive it from whatever they are using. The page covers what that means in practice: a table that accepts fixtures nobody has drawn yet, an SDK rather than a fixed application, and a calibration record that lets a new group verify the machine instead of trusting it.
The performance figure that matters also changes. Peak acceleration decides very little in a laboratory. Repeatability decides almost everything, because a result is only a result if the input can be produced again — by a different person, after a payload change, next year. That is a different acceptance test from a peak-performance one, and it is worth writing before the machine is built.
Writing a grant or a tender for equipment that has to last? Open the research platform page →
Application 03 · The parent page for every simulator project
Motion platforms for simulators, supplied under a device somebody else owns
An integrator, an OEM or a training provider owns the simulator. What they need underneath it is a moving machine, its control, and an interface contract they can write software against. That is a platform supply, and its boundaries are worth drawing before anything is built.
The idea this page is built around is that the motion is not the cue. A vehicle in a model can accelerate forever; a platform has finite travel. What converts one into the other is the washout filter, and it is the filter rather than the stroke that people are usually reacting to when they say a simulator feels wrong. A longer machine with a badly set filter feels worse than a shorter one with a good one, which is why the cue set belongs in the specification alongside the envelope.
This is also where the scope boundary is drawn for the two specialist pages beneath it. The platform, its controller and the interface are supplied; the cockpit or buck, the visual system, the content, the host model and the qualification of the finished device are not. Where a project has not yet decided whether it is a flight device or a driving device — or is neither — this is the page to start on.
A simulator that feels wrong is usually a filter problem, not a stroke problem.
Building a simulator and buying only the motion? Open the simulator platform page →
Application 04 · A specialist child of the simulator page
Flight simulator motion platforms, where the motion is examined rather than admired
Flight training devices are the one application where somebody outside the project measures the motion and writes down what they found. That single fact reorganises the whole specification.
The page is built around the objective motion cueing test, which was made a requirement for initial evaluations of flight simulation training devices after 31 May 2016. It runs ten conditions — six direct-transfer and four cross-coupling — and measures the platform’s frequency response against the aeroplane model driving it. What matters commercially is that Part 60 assigns no pass tolerance to it: the tolerance is proposed in the qualification test guide and ICAO 9625 Edition 4 supplies fidelity boundaries. So the number your device is judged against is partly a number your programme proposes, and that has to be understood before hardware is ordered.
There is a geometric consequence the page works through as well. A pilot’s head sits about a metre above the centre of rotation, so a single degree of roll moves it roughly 17 mm sideways. That lever arm is why motion reference point placement is a specification item and not a detail.
Programme has an evaluation date and a motion system still to buy? Open the flight simulator page →
Application 05 · A specialist child of the simulator page
Driving simulator motion platforms, where gravity does the work the stroke cannot
Driver-in-the-loop work asks for something a platform cannot literally provide: a sustained longitudinal acceleration, held for as long as the vehicle would hold it. The page is mostly about how that is solved and what it costs.
The arithmetic is worth seeing before a specification is written. A 0.5 g braking event needs a tilt of about 26.6° to reproduce the sustained part through gravity, because the tilt angle follows the ratio of acceleration to gravity. Tilt onset has to stay below roughly 3° per second or the driver perceives the rotation instead of the deceleration — and at that rate, reaching 26.6° takes nearly nine seconds. A real braking event from motorway speed is over in about six.
So the cue is split: the stroke supplies the onset, gravity supplies what remains, and the handover between them is where a driving simulator is won or lost. That is a tuning and specification problem more than a hardware one, which is why the page spends its length on the cue set, the loop and the timing rather than on travel figures.
Stroke buys the onset. Tilt buys the rest. The handover is the engineering.
Running a DIL programme with a buck already designed? Open the driving simulator page →
Application 06 · The one that runs in reverse
Motion compensation platforms, judged on the motion that is left
Everything else in this column is bought to produce motion. This one is bought to remove it: a base moves, and a payload on top of it is supposed to move as little as possible.
Because the platform is chasing a disturbance rather than following a command, latency stops being a quality metric and becomes the whole problem. The residual left after a compensation attempt is set by how far the correction has fallen behind, and the relationship is unforgiving: at a delay of one sixth of a period, the correction contributes exactly as much motion as it removes. Past that point the machine is making things worse than doing nothing.
Run the other way, it gives a design rule you can use before anyone quotes hardware. Cutting a disturbance by 90 % needs the whole loop — sensor, transport, solution and servo response — inside roughly 32 ms at 0.5 Hz, and inside about 8 ms at 2 Hz. That is why the page asks for the base motion and the sensor before it asks about the platform: a mechanism cannot fix a latency budget that does not close.
Compensation is a latency problem wearing a mechanism. Feasibility comes before hardware.
Have a recording of the base motion you need removed? Open the motion compensation page →
True on all six, whatever the job is
Six different applications, three identical questions
The pages above differ in what they have to prove. Underneath, every one of them starts from the same three questions, and a project that answers these can be quoted whichever page it lands on.
The complete moving assembly
Not the payload — everything above the actuators, including the table, the fixture or cabin, the occupants and anything they carry. Three numbers describe it: mass, centre-of-gravity height above the table, and inertia about each axis. Mass alone sizes a machine badly, because height and inertia decide the moment the actuators have to supply.
The interface contract
What your software sends, in which units and frame, at what rate; which way is positive on every axis; what the machine does on a fault, at a limit, and when your software stops talking to it. Agreed and signed before manufacture, because it is the one document two companies argue about at commissioning.
The loaded acceptance
Every application is accepted the same way: ballast built to your mass, centre-of-gravity height and inertia rather than stacked as flat plates, the agreed profiles run, and the channels logged and handed over as files. What differs between the six is which profiles and what counts as a pass.
What is supplied on every one of the six, and what is not
The supply is the moving machine and its control: platform, actuators, drives, controller and cabinet as one delivered unit, with the API and SDK, the calibration data set, the interface documents, the structural load case and supply sheet, the loaded acceptance test, and the documentation serialised to the machine.
What stays with whoever owns the work is the article on the table and whatever it models or measures — the cockpit or buck, the visual system, the training content, the test method, the vehicle model, the research programme, the vessel, and the safety case for the installed system. That boundary does not move between applications. It is the same on a laboratory rig and on a thirty-tonne motion base, and it is written into the scope before an order rather than discovered at commissioning.
Need the engineering behind these three rather than the summary? Open the Custom & OEM column →
Once the application is settled
Where this column hands over
Applications decide what the machine has to do. Three other columns decide what it is, how it is engineered and what it costs.
Range
The platforms themselves
Published envelopes for each machine family. The page to open once the application is settled and the question turns into which mechanism serves it at your load.
Custom & OEM
When no standard machine fits
Four routes from a configured catalogue machine to a new design, what each one re-opens in analysis and testing, and the twelve lines a quotation needs.
Trade-off
Payload, stroke and acceleration
The three targets in an application specification that pull against each other, worked in numbers. Read it before writing the specification, not while defending it.
Vocabulary
Motion platform glossary
One definition each for the terms these six pages use most: motion reference point, gross moving load, washout, cue set, forward and inverse kinematics.
If the machine is settled and only the software is in question, the controller and API page is the right next step; if a building is involved, the integration page lists what a site has to provide.
Asked while choosing between these six
Motion platform application FAQ
My project sits between two of these pages. Which one do I use?
Use the one that owns your acceptance test, because that is the thing you will eventually argue about. A test rig that also has to feel plausible to an operator is still industrial testing, since the pass criterion is a reproduced trajectory and the feel is a preference. A driving simulator that also has to replay a recorded road profile is still a driving simulator, since a driver is the thing being convinced. Where the two genuinely carry equal weight — which happens most often between research and industrial testing — send the objective and one representative motion file, and the reply will name the page and say why the other one was not chosen.
Does the application decide how many axes I need?
It decides which axes have to be real, which is a more useful question. A compensation duty against a rolling deck may need only the two rotations and heave, and three axes done well will beat six done cheaply. A driving simulator needs longitudinal travel and pitch specifically, because that is where the braking cue is built. An industrial test that reproduces a recorded six-degree-of-freedom trajectory needs all six or it is reproducing something else. Starting from the axis count and working backwards is how projects end up buying motion they never command.
Can one platform serve two applications?
Often, and it is worth asking for explicitly rather than hoping. A research platform used for both experiments and component testing is a common and sensible combination, because both want repeatability and a documented input. A machine expected to serve both a simulator and a compensation duty is a harder case, because one is tuned around perception and the other around latency and residual motion, and the acceptance tests point in opposite directions. Where a dual use is intended, both duties belong in the specification from the start — retrofitting the second one later usually means re-opening the sizing.
Why is this column organised by engineering objective instead of by industry?
Because the industry label predicts the machine badly. Two aerospace projects can need opposite platforms — a component test rig and a training device share almost nothing except the sector on the purchase order. Meanwhile a marine compensation duty and a laboratory compensation duty are nearly the same engineering problem in different rooms. Sorting by what has to be convinced puts projects that share an acceptance test on the same page, which is what makes each page able to state its inputs and its proof concretely instead of listing industries.
What is supplied on every one of these applications, and what is not?
Supplied: the platform, actuators, drives, controller and cabinet as one machine, the API and SDK, the calibration data set, the payload and control interface documents, the structural load case and electrical supply sheet, the loaded acceptance test with its logged channels, and documentation serialised to the machine. Not supplied, on any of the six: the article on the table and whatever it models or measures — cockpit or buck, visual system, training content, test method, vehicle model, research programme, vessel — and the safety case for the installed system. That boundary is identical across the column and it is written into the scope before an order.
Do you supply complete simulators or complete test systems?
No, and it is the most common misunderstanding in this column. We build the moving machine and the software that drives it, and we supply it as a subsystem to whoever owns the device. For a simulator that means the integrator keeps the cockpit, the visuals, the model and the qualification; for a test system it means the laboratory keeps the method, the instrumentation and any accreditation. There are two reasons for the boundary rather than one: those parts are somebody else's specialism, and a supplier who claims all of it usually cannot be held to any of it.
How do I compare platforms across suppliers when every application quotes differently?
Normalise three things before comparing any number. First, whether a payload figure means gross moving load — table, fixture and article — or only the article, and at what centre-of-gravity height it was sized. Second, whether an acceleration is a single axis alone or a combined case with the other axes active. Third, whether the acceptance is run at load with ballast built to your centre-of-gravity height, or run empty. Those three account for most of the difference between quotations that appear to describe the same machine, and each of the six pages states its own figures with the conditions attached for exactly that reason.
What do you need from me before you can even name the application?
Three things, and none of them requires a decision about a machine. What the motion has to achieve, in one paragraph and in the physical terms you care about. The complete moving assembly — mass, centre-of-gravity height and inertia, or the drawing they can be taken from. And one representative motion file, recording or specification of the motion itself. With those, the first reply names the application page, the two or three inputs still missing, and what each of them would change. Without them, any answer is a guess wearing a model number.
Commercial questions — lead time, payment terms, warranty, shipping and installation — are answered in the full FAQ.
Send the job. The first reply names the application, not a model.
What the motion has to achieve, the complete moving assembly and one representative motion file are enough. If the right answer is a standard machine and a shorter conversation, that is what comes back.
What comes back first
- Which of the six application pages owns your project, and why not the neighbouring one
- What that page treats as proof, translated into an acceptance you could put in a contract
- The two or three inputs still missing, and what each one would change
- Whether this looks like a standard machine, a configured one or a designed one