Question Are 4 legged motion platforms actually just "heave" simulators?

Woth their low point of rotation, are 4 legged motion platforms like the SFX100, PT-A, D-Box,.. just glorified "heave simulators"?

From my understanding all rotational (roll, pitch, yaw) degrees of freedom should be anchored to the 3 dimensional center of mass (gravity?) and transitional (heave, sway, surge) degrees of freedom should move along the planes of XYZ.

What I conclude is that those style of motion platforms are actually only to truthfully reproduce "heave" because of their low mounting point and center of gravity and all other experienced degrees of freedom (pitch, roll) are just side effects that are "off" what they actually should be.

Discuss.

(this isn't meant as trolling, it's a genuine question/concern of me after researching for a prolonged time now on how to properly set up a 3 legged 3dof platform)

Yeah, but that's not what I was after.

Imho every degree of freedom should move independently from the others. That's just not the case with dbox style platform movers bolted on a rigid base. Hence my conclusion the only DoF they can simulate properly without side effects is heave, a combination of more than 1dof (pitch, roll, heave) isn't possible with 4 rigid mounting points as it's over-restricted. It can do roll or pitch on it's own but will introduce sway or surge as a side-effect (same imho goes for a 3 legged platform with a mounting point below CoG, but at least your platform isn't overconstrained and can do all 3dof at the same time without interference.

For a true 3dof without side effects imho you'd have to have the pivot at center of mass - and you should be as near as possible at that center of mass. Roll and Pitch would need to be realized by a pivot point right in the centerpoint to minimize side effects of sway and surge.

a 4 legged sim is only a 3dof - not a 4dof as some people think

they are often used as a way to develop a static sim into a pegleg motion sim.

it makes sense to use 3 actuators not 4 from an engineering point of view

however you can make a stable peg leg with good motion control






there is some discussion about it here

"4x drivers for 3DoF design to me is a "faulty design". Searching through all published and research papers and robotics & motion platform industries just could not find a single such 4x for 3DoF design. Similarly, for 6DoF, if one adds a 7th actuator, then will not work with 6DoF. It constrains the motion, and no longer be able to produce 6 degree freedom motion. Today, all motion platforms designed as n x drivers for nDoF config. I never have seen such n+1 drivers for n DOF (except for D-Box). I found only the most expensive sim NADS/Lexus/Toyota uses 4x actuators on the corners, but their purpose is for vibrations only, not for motion. The max vibration amplitude limits under < 0.2 inch though:

https://www.researchgate.net/public...TY_OF_IOWA_A_TOOL_FOR_DRIVING_SAFETY_RESEARCH

Sadly, most people don't realize such fact, thinking because there are 4 tires on a car therefore 4x must be right and better. No, it is not. It is still 3DoF nothing more, no matter what you try, the chassis is a rigid body in this case still does 3DoF. By archiving so, the 3x config will offer correct 3DoF motion while 4x config dose not. To me it is worst to think of putting 4 actuators on 4 corners. If I turn up the motion intensity I will get floating feel and cannot feel roll/pitch coherently while high speed cornering. If I turn down the motion all the way, then I got "tactile" effect only. I do recommend others to go with 3x or 2x. "

I understand its hard to go from a static sim to motion and a step up to 6dof is quite an expense

the pegleg sims I have tried have all had very good resolution on the heave detail using stepper motors

What did I do wrong that this is ending up to be a 3leg vs 4leg discussion?

I'm actually trying to dismiss both of them, when mounted on a rigid platform with a pivot far below CoG.

Hoping for someone to point out a flaw in my thinking proving me wrong.

I think about this a lot. So much so that I built a pegleg with 4 actuators, a pegleg with 3 actuators, an inverted pegleg with a base and the center of rotation at the center of mass - on three actuators and I am now building one with 4 actuators. Plus I went way down the g seat path and the g helmet path.

These sims we build are always delivering partial degrees of freedom. It would be impossible to do it perfectly blah blah blah.

They dont simulate heave. Only high frequency heave. So even that is flawed.

The pitch and roll axis on a pegleg sim is pretty shit for pitch and roll, but its great for high frequency surge and sway. This is because of what you pointed out - the center of rotation is on the damn floor. When you rotate from that point for small angles it is really more like translating side to side or front to back. It works really well actually, but only for high frequency small motions. I believe this is the biggest reason so many people say less is more in vr (thats another topic all together). The pegleg is such a great design because of its high value. For what you build you get a lot back.

Its not without false cues though. Using pitch and roll to cue lower frequency surge and sway (like a turn) is pretty bad. You get really bad false cues from the center of rotation being so low.

If you build a sim with the center of rotation at the center of mass then you dont get shit for surge and sway but you get pretty darn good pitch and roll(its pretty damn cool to feel steep pitch showing you the elevation your climbing on the Nürburgring for example). I think a lot of people, if they had to choose, would pick feeling high frequency surge and sway over better pitch and roll.

I think I agree with you that a pegleg sim is limited but I don't think I agree with these words. You can certainly program the sim to have independent degrees of freedom. Most people blend surge with pitch and sway with roll so they can get more data from the game with what limited motion they can reproduce.

I think it is unfortunate that we don't have a better vocabulary around the physics of sims. There is so much confusion out there - especially with the term dof. Hopefully the more we discuss this stuff the better we can articulate things, and clear up some of the fog.
Please let me know what you think about my response. I do enjoy thinking about this stuff.

Thanks, this is the kind of discussion I was after.

I have to admit I like to think about it more than I have time to actually build stuff these days.

One thing that's always eluded me was how current motion sims (software) handle the driver being seated offset in a car (like in a GT3 or LMP2 vs a single seater). Do they actually account for this?
Imagine running a steep curb on the right hand side when seated left in an LMP2. I'd argue that a pegleg sim can replicate that pretty well potentially if you can sort out the sway/roll correlation. But does it differ if you run down that curb being seated to the left (small heave) vs running down the same curb seated on the right hand side of the car (huge heave)? Does it matter in reality?

If you'd build a 3DOF sim - what would you prefer out of those 2:
1) Heave + Sway + Surge
2) Roll + Pitch + Yaw

I personally tend to say I'd probably prefer 1) over 2), but maybe I'm biased as a former seatmover owner which imho can do sway/surge not too bad for my liking (I never minded the changing distances between pedals, wheel and my body. Used to run it hard with very low travel and ot gave me a convincing sensation). I also dialed out pitch and roll almost completely, didn't find those really useful in a seatmover. It was all about muscle activation for me. I know it's inherently wrong by all kinds of measures, but it felt great to me.

One thing that's always eluded me was how current motion sims (software) handle the driver being seated offset in a car (like in a GT3 or LMP2 vs a single seater). Do they actually account for this?
Imagine running a steep curb on the right hand side when seated left in an LMP2. I'd argue that a pegleg sim can replicate that pretty well potentially if you can sort out the sway/roll correlation. But does it differ if you run down that curb being seated to the left (small heave) vs running down the same curb seated on the right hand side of the car (huge heave)? Does it matter in reality?

some software simulates forces with an offset driver position - you may be able to adjust cog to suit a car if that doesn't work

I don't know for sure but my strong intuition is that they output the physics of the vehicle not the driver. But it would certainly be up to the game designer. In reality I don't think we humans would really be able to tell in a racing environment. Too chaotic. We really get pretty course data through our vestibular and skeletal system.

Oh its a no brainer. Option 1. I have in fact experienced 2 both with a floor COR and a COG COR (haha) and its neat, like hearing a song backwards. But you would never prefer it.

Funny thing is I am building a chassis that only has heave, roll, pitch, and full 360 continuous yaw. I'm going to be missing out big time on Surge and Sway. But without a stewart platform I think this is the best way.
I will provide surge and sway through better methods like g belt, g seat, and g helmet.

I know what you mean. I find that there are lots of great cues we can make to deliver data to the body that dont replicate exactly how a car behaves.

Check out the seat bucket on in this video. The sensation is freaking amazing for surge and extreamely relatable to real life. It puts pressure on your lower back like you are forced backwards in your seat. But it also gives you a hell of a jolt for gear shifts etc.





I think getting the data(vehicle physics) into the body should be all of our goal not realism. Of course it would be nice and we should strive to get the data in through a mechanism that we can relate to from real life.

The thing that bugs me the most - and it really really bugs me, is how to accurately deliver acceleration data (say through pressure) into two different bodies. IE, if I get a friend in the sim it feels totally different for him as it does for me because he is a different shape and a different level of squish. The corollary to this is how people scale their sim intensity or travel up or down for different cars. Seems to me we should be able to build a sim that nails any given game engine and the game designers show us the differences in the cars. Not doing this makes it hard to feel the glorious differences between vehicles and harder still to get the same sensations as someone else.

Yeah, that G-Seat looks pretty neat, but won't be in my books for some time due to space and noise constraints.

What I did think about doing though was putting the seat on rails and let an actuator push/pull on it for surge (seatmover style, but without the tilting) and have a passive belt tensioning mechanism behind the seat. To me that sounds like best of 2 words in theory. I have also been playing a thought in my head for quite some time: hook the surge effect up to a torque controlled actuator (still have a few Argon and IONI drives around with torque control) so it may pressure you against / pull you away from the static wheels and pedals without actually moving too much.

About the G-Seat conundrum of different bodies: have you thought about compensating for body shape and size maybe with some foam inserts to get the driver in the right positions for the pressure points?

My system uses a push/pull slider for surge working independently on top of the motion platform as you have described. It gives a very convincing effect for acceleration and deceleration forces by moving the seat fractionally closer and away from the pedals and wheel. The total travel is only around 5cm and so I use active belt tensioning as this allows greater tension on the belts without having the slide movement to feel unrealistic. I made mine with an inexpensive but powerful DC servo (~120kg/cm) but if I was designing it again, I think I would use a screw style actuator setup as I occasionally put enough force when braking to over-strain the servo.

Have you considered using airbags ?

Although a bit different in design then most, I have a three legged platform were I can adjust were the Roll center is in relation to the simulation frame (just under the seat at the moment). If required, I can drop the frame another 36cm, which would move the roll center into the middle of my back. Looking to add in some G systems in the future, but am happy in how it performs at the moment.

If you are in a real car, when the car roll or pitch whether because of road surface or suspension action, the center of rotation will be at the floor level. That means a small amount of sway & surge will also be there. Much like the 4 legged sim.

I believe most of racing game will output the roll & pitch of the car rather than the roll & pitch of in game driver. So its better to have a motion sim that replicate the physical of a car. Example : center of rotation under the driver instead of at the driver's center of gravity.

4 actuators will not over constraint a 3dof system if done correctly. The program must make sure that all 4 feets will always be on the same plane. The 4th actuator also has a purpose in engineering point of view, that is to give stability without the need to bolt the feet to the ground.

With 3 actuators you need to bolt the feets to ground or have a large enough foot print or else stability will be an issue.

I have a 4 legged sim and I make the arduino program for it. At one time I made the option to have pitch center of rotation behind the driver like a real car do and roll center of rotation to be selectable either at center, left or right. You can do this if you have heave ability. But I scrapped it because my actuator only have 75mm of travel and that option will further limit roll & pitch angle. Also because it didn't make much different.

Consider a 4 legged motion platforms is more a theoretical error free way to simulate the chassis or suspension in a car racing simulation.instead of a moving device, assumed suspension telemetry values are used.The reference point in that case is road surface.

Simulate tilt and roll relative to zero ground is imo. especially for roll not convincing because of missing the dynamic forces in real, mostly reacting in a false cue.

An additional advantage, monitors must not necessary mounted to the rig.

Anyway up an down hill is already visual simulated in monitors, missing is only a tiny neglectable component of the drivers g-force e.g. realistic 12% slope ~ arctan 0,12 is only ~ 7 degree.

Basically only a dynamical real position mapping to a static sim called x-dof is not constructive for realistic car simulation.The principle should be to simulate forces not position.

Is that you agreeing with me that it's more or less just a heave simulator or disagreeing?

its or should be a heave with pitch and roll referenced to road surface in contrast of sea level , or in other words movement of the suspension.Consider also its a physical mismatch mapping vert. acceleration to heave position.

a peg leg 4 legged sim is only 3dof

you can get the same result with 3 actuators not 4

with respect to surge normally the longitudinal acceleration is mapped to a position value.

Meaning likewise getting a physical dimension error from m/sec2 to m, resulting partially in false cues or as a kind of swinging effect if sitting on it..

This could be imo. eliminated if wheels and pedals in front are put on rails, only deceleration is processed to get forces under braking to drivers hands. Mapping from deceleration to force, F=m*a, F~a