Simulator designs general discussion

*This thread is a continuation of a discussion, which started here: post25932.html#p25932 *

I actually don't see what you mean, the movement are more like a simforcegt and joyrider combined.
so it sounds like it does more than the known movements on this forum.
the swinging movement plus the bumps is the goal for me to.
I could be wrong, but could you point out the movements that are really off?
Not that i disagree, I just don't see it.
And about the below seat pivot isn't six dof platform doing just that?
I'm also more interested in the how they did it, has anyone got an idea?

Re: new simulator : JoyRide

2 solutions , i think he use actiator or motor like me to move the simulator.

Re: new simulator : JoyRide

All well and good but in that respect every 'simulator' on this site falls under the same category. Unless you win the lottery you're never going to be able to put together a 'simulator' that's going to mimic these forces. And of course, if you won the lottery you'd probably be driving the real thing.

The simulators here work mainly on the principle of fooling the brain into thinking you are being subjected to similar, note I say similar, forces. Some do it better than others but it's not fair to pick any particular 'simulator' for the reasons you have stated.

Just my thoughts ...........

Re: new simulator : JoyRide

R-eng is correct in the sense that this type of simulator construction and geometry offers poor reproduction of G-forces.
It is still a simulator, but not an accurate one.

wimpo:
Just think about what happens in a braking scenario in the Atomic (or same pivot-under-seat joyride design). The whole thing tilts FORWARD, which is all and good, but think about where your head is going and it's trajectory in space. It is also going FORWARD -> this is to simulate acceleration forces and NOT braking forces!
Now consider when you accelerate. The seat tilts backwards, and your head is also moving backwards -> again, this is a force felt under braking, and NOT acceleration.

Same thing applies to side movements. The head feels OPPOSITE G-forces than what your body feels from tilting. Since most of your balancing mechanisms are in your head, this geometry will produce inaccurate sense of forces.

By placing the pivot point above the head, the direction of these forces will be correct. When you tilt backwards your head will move forward, when you tilt forward, head will move back -> proper forces are applied to the whole body.

This doesn't apply to seatmovers, where the wheel and pedals are stationary! These simulators work on completely different principles to simulate forces.

6DOF platforms are not tied to only 2 axis and their Center of Rotation point can be adjusted through the control algorithm. Their mechanical pivot points have nothing to do with the pivot point of the upper platform.

riton: I will bet you money there are 2 actuators under the Atomic, very similar (if not the same) to SCN5's.

Re: new simulator : JoyRide

Well in my car when I accelerate rapidly my head moves backwards, it definitely doesn't move forward. Also, when I had to make an emergency stop on the motorway yesterday my head definitely moved forward - along with everything else in the car I might add.

Re: new simulator : JoyRide

I think Frak is trying to refer to the difference between head movements and bodily movements that are picked up by your inner ear. Yes, your head went backwards when you accelerated, but your whole car went forward (not backward) which effects your inner ear, sense of balance, etc..

I think there are some valid arguments for either approach. IMHO, the closest we can reasonably achieve today is to place a 2DOF seat (to represent G-Forces) atop a 6DOF base. This eliminates alot of the motion cue conflicts that exist in most sim designs that exist day. I have been working on this for some time now. It sounds alot simpler than it is

Re: new simulator : JoyRide

Knowing your products Bernard I look forward to seeing this at some point in the future !!

Re: new simulator : JoyRide

i work to put a FREX design under my simulator, i think it's a good solution, there are no best simulator...
i think the best simulator is a mix with different simulator.
we need alls solution to simulate , and have good feeling.

frex design+ simulator like me ( name ? RNR)+vibration, basse pump...

Re: new simulator : JoyRide

I have covered this subject a couple of times before, but I will explain once more...on one movement only.

@Frakk, you are nearly correct.

Braking. In a real car, when you brake, the chassis will slow down. Your head will not, or rather not as quickly. It is attached flexibly to your torso via the neck. When you brake the car, the body slows with the chassis because it is strapped to the seat, anchored to the chassis. Your head will pivot forward due to physics. A body will want to continue in motion until that motion is redirected or restrained, as is the case with the neck (just like a tether).

Joyrider systems with pivots anywhere below the auditory and vestibular pathways, will always, always move the head in the wrong direction even though part of the torso (below the pivot) is moved correctly. The joyrider pitches forward to simulate G's pushing your body forward. That part of your body above the pivot point will actually be moving backwards in the seat on initial surge (= acceleration force). Only when the motion is completed will the G force be correct. By that time the brain has already locked on to the motion cue, The wrong cue. Furthermore, your head wants to stay still, lagging the movement of the torso, until the neck pulls it forward. This is the same motion as in acceleration. Yet all the time you are attempting to produce deceleration motion cues. Physics does not lie. This form of joyrider is an arcade device, not a simulator.

from Wiki--
Simulation; is the imitation of some real thing, state of affairs, or process. The act of simulating something generally entails representing certain key characteristics or behaviours of a selected physical or abstract system.

read carefully...the imitation of some real thing.

All other forces applied with this form of joyrider and frex systems, is similarly wrong. Where the frex deviates is in simulating bumps and verticle upsets; There it can be accurate, but not in lateral and longitudinal G forces.

Sorry, but that is the physics explanation, and the truth. That does not mean you cannot enjoy these systems, only that they are not simulators of the real process, racing/street vehicles. Period.

If true simulation were so easy to accomplish, why is it so difficult and expensive to get it right???
It would have been done a hundred years ago, and would only cost about <$100 to achieve today. McLaren would not have devoted over 10 years and over $10,000,000.00 to the subject.
Get real. These systems are easier and cheaper to construct/operate. The more balanced they are, the smaller the motors and forces needed to move them. The simpler and weaker the structure can be. Part of your body moves correctly, the other incorrectly...guess which is incorrect?

R-eng

Re: new simulator : JoyRide

@EgoExpress - Can we move these off-topic posts to aa new thread about simulator design theory?


No doubt that a large amount of R&D spending is required to produce something that provides a significant number of accurate cues!

Your intrepretation of the physics is interesting. There are many such intrepretations resulting from various studies. In the end, it's about making the best compromises given a target price range and set of vehicles to simulate.

As I mentioned before, there are valid arguments to just about any approach and some element of truth to them all. The difficulty of course is that they can't all be right and therefore many of them contain half-truths either via honest mistake or via intentional market positioning and deception. Most of us who have been through hundreds of test subjects can understand how easy it is to get bad feedback so I'm not casting stones at anyone's theory.

When I first started toying with sims, I built several different styles for comparisons sake and armed myself with dozens of whitepapers on the subject. Most of the whitepapers and classroom theories directly contradicted each other and agreed on very few common elements. Some of them frankly drew completely illogical conclusions.

Since the combined opinion of the experts was All dogs are animals -- All cats are animals -- Therefore, all dogs are cats, I obviously needed to conduct my own research.

I have found the 2DOF seat mover sims to do a great job of representing both lateral and longitudinal G-Forces (otherwise I would have built something different). The key is the proper motion profile. We control the speed, acceleration and amount of each movement. However, it is important to understand that the G-Force movements directly contradict roll and pitch movements which it why I am taking things to the next level with a 6DOF platform + 2DOF seat. This is also a point of contention when having these types of conversations because many experts do not seperate these concepts. G-Force movements are entirely different from vehicle Yaw, Pitch and Roll!

I think military simulation researcher Phillip Denne said it best when he said:

I have found all of these points to be true. I have also found that sim racers desire an entirely different feel than race car drivers. Sim racers are looking for a competitive advantage via cues and actual drivers are trying to make their sim feel as close to the real vehicle as possible. Some of the elements desired by sim racers violate a few fundementals, but again, their goal is first to win online races and second to have as realistic an experience as possible. This doubles my work because I have to create multiple sets of motion profiles for each.

I have seen many 2DOF seat motion profiles that suggest lateral g-forces / roll can be combined and that longitudinal g-forces / pitch can be combined. I have combined these and while it does improve a few cues (such as placing a wheel on a curb), it caused more negative cues than positive.

I would like to improve this conflict area, which is the reasoning behind my 6DOF + 2DOF (btw - this concept is not original, the military has been doing this for a very long time and so has Phillip Denne).

I prefer to have roll (such as the grade of a NASCAR corner) in the 6DOF independently of the the lateral g-forces (in the seat) such that the entire vehicle is at the same angle it would be on the track, but the seat is representng the g-force fluctuations.

So there it is, yet another opinion on the subject for what it's worth.

:

Re: new simulator : JoyRide

So is this a good explanation?


The forces applied on your head is Inertia.
And the movement forces are Motion.

The one one the right won't give a good motion I think.
because the head wont be moving to much, more like tilting your head a little bit.

Re: new simulator : JoyRide

I agree, this is getting too much off topic now, a new thread would be very nice for the discussion!

We have to make a very clear distinction between simulators that ONLY MOVE THE SEAT vs. simulators that MOVE WHEELS and PEDALS with the driver.
For the sake of simplicity, let's not get into moving/not moving the screens or displays and the visual aspect of simulation.

I wouldn't draw seat movers into the pivot below the head fault discussion, because you cannot compare apples to oranges.

Bernard, you are absolutely right with the explanation and I understand you are looking at this from the seatmover point of view.
Your sims rely on totally different principles to represent motion cues and force effects than a joyrider type sim, or the Atomic for example.
Seatmovers have a fairly limited motion range and use the changing distance and position between driver and wheel/pedal to simulate the forces mainly on your MUSCLES.
There is relatively small actual G-Forces applied, and sustained to the body, therefore the reverse effects of inertia on the head won't be as noticeable, if at all.

Now considering a simulator in which everything moves with relative to the seat and driver.
This type of construction cannot produce muscle tension effects, as the position and distance between wheel/pedal is fixed. These simulators rely on tilting and shaking to represent actual G-Forces.
Because there is a greater range of motion, the actual force simulated on the body can be greater than in a seat mover. On the down side again, they cannot produce any muscle tension effects.
In this type of simulator the reverse effects of inertia will be much more noticeable due to the increased tilt and the lack of other motion cues/stimulated forces. The lower we put the pivot below the head (increase radius), the circumference of the circle will also grow, hence your head will move more distance for every degree of tilt. This in turn will cause a reverse force due to the inertia in the beginning and at the end of the tilting motion.

Nice diagrams wimpo, I would have included one that shows what happens when you put it above the head, but you get the idea.

Re: new simulator : JoyRide

Nice illustration Wimpo. Remember that in each design, the motion profile can offset some of the design issues. For example, when a lateral movement is made, you can accelerate the actuator SLIGHTLY more quickly at first and slower toward the end of the movement in order to get the intertia + some sustained motion.

Of course this is where every simulator falls short. We don't have an unlimited amount of actuator travel to create a sustained motion and that is one of the reasons I prefer to focus on muscle tension aspects as well. They are sustainable and fortunately mix well with the other aspects, but only in moderation.

So in short, you can't create the actual g-forces of the vehicle and there isn't enough actuator travel to sustain the motion cues. The best you can do is to create accurately scaled movements within the bounds of the accurate muscle tension range IMHO. When you want more than this, my opinion is that you need to place a 6DOF base beneath the entire sim.

Re: new simulator : JoyRide

The idea with simulators is that the visual sustains the cues. But the motion onset cue should be accurate in the first place, with the rest of the scaled movements continuing to be accurately and proportionately conveyed. Otherwise the mind gets confused when relating to reality, and the whole experience becomes muddled.

Perfecting dynamics with Formula 1 principles
Since 1997, we have been developing and using an immensely powerful driving simulator to predict and refine dynamics, safety and perfornance. Widely regarded as the most sophisticated in the industry, details of our simulator remain a closely guarded secret, even within the McLaren Technology Centre.

From Formula 1 to road
Our Racing team rely heavily on McLaren's sophisticated simulator for the testing and refinement of their cars...

--McLaren.com

Well I was wrong, over 13 years now. And the cost is wrong too! at a cost estimated to be above £20m

From the snippets of info I have been able to gather, the McLaren simulator is a form of 6.DOF Stewart plaform which itself moves in 2 directions on a horizontal plane structure about the size of a basketball court. Kind of like a very large Ball Racing Development unit. Toyota have something similar to this, and NASA has a verticle version also. This is one way to sustain near real levels of G forces, since they are really short lived and transient in Formula 1.

R-eng

Re: new simulator : JoyRide

There is a post somewhere in these forums with vids of some of the top auto manufactures' simulators. I don't recall anyting able to move about a basketball court sized area. That would be interesting to try, although it would still not be able to sustain Formula 1 level G-Forces in any corner longer than a basketball court unless it was circling the baskbetball court sized area, which would present a whole different series of problems.

If we stop at simple cues and believe that visuals are enough to sustain the cue / immersion, then we're done and achieved the holy grail of simulation when the first Frex was made and provided simple onset cues with muscle tension capability and a heavy reliance on visuals. I myself would prefer to continue innovating. I think it would be boring to be involved in an industry or hobby that has already fully evolved.

There are several theories on how to handle sustained g-forces but I think as an industry, we have not yet arrived at an optimal answer, although I don't think we're far away.

What do you think of extending a cue? Suppose you could hve 4 times the actuator travel that you have today. Do you perceive any value in extending the cue or do you find an inch or two of travel to create the initial cue to be enough?

In few words I'd explain it this way:

-Pivot over head (POH): This concept applies correct g-forces to the driver instantly.

-Pivot under body (PUB): This concept requires the body (mainly the head) to be tilt in the wrong direction initially, until the body reaches the adequate position to experience the correct g-forces (regarding vehicle). This is definitely a con.

-Pivot at body center (PBC): This is actually a compromise between PUB and POH. Though it adds some rotational forces to the g-forces applied to the body. But entering/leaving the simulator may require some athletic skills, depending on the design.

-Seatmovers are a special case: though they are PUB concepts, the wrong movements arent that much noticeable due to their restricted movement range. As well they mainly dont simulate g-forces, but the sensation in the arms while driving, due to changes in body arm lenght and their angle to the body. The advantage of seatmovers is definatly their low mass, which lets them move the driver amazingly fast.

However, I wouldnt go that far to disqualify PUB concepts. It depends on what you want to achieve, and probably on your budget as well.
I think POH concepts are likely more expensive and difficult to build, and may need stronger mechancis/actuators.
And it depends on what you want to simulate as well.
A PUB design suits IMO quite well for slow operating simulators like a Cessna or a Boeing simulator for example, which doesnt require movements with light speed, and therefor the wrong movements arent experienced that much.

IMO every design has its advantages, flaws and application. Everyone has to decide for himself which concept fits the most to his needs. There is no one size fits all simulator concept.

Regards

Re: new simulator : JoyRide

I must narrow my scope. I use a simulator for car racing. It would be the holy grail if visuals could truly sustain a motion cue...but really it cannot, depending on duration.

Racing is highly transient in real life, at least for good drivers. When you look at the data traces of lateral G motions, whether from real sources or the sim, the driver is constantly adjusting between OS/US to maintain the vehicle at the ragged edge of adhesion. If the simulator is not set up to be soft, then it will be constantly in motion within the turn. There is nothing to sustain really. The extreme brief modes between adjustments to steering corrections will indeed by long enough to allow the brain to be fooled, and so that is why the visual can sustain the cue, to a degree.

Is it enough? I'm sure it depends upon the person and the immersion level attained within the simulator environment. And whether the cue onset was correctly portrayed in the first place.

This doesn't even touch on human sensory input latency, or the latency from the controls and simulator in the first place. We may just be reacting to yesterdays news with some simulators, which I have clearly seen within some videos in this forum.

R-eng

That's an interesting point about the changing G-Forces when on the ragged edge of traction, but I'm not sure that it's entirely accurate. I have studied many data logs and in each of these cases, there is a sustained G-Force. The line may become jagged as the driver flirts with traction, but never reaches 0 (or anything close to it) while still in the corner unless traction is lost completely.

I couldn't agree more about the effect varying by person. I've been trying to narrow doen the variables for this, but overall, I don't anyone has scratched the surface of this yet.

Again, no doubt that many sims have significant delay. The ones seen here aren't a reflection of X-Sim, but of the chosen hardware or motion profile.

I fully agree with this.

Regarding the fluctuations, positive and negative moving G forces are completely riding atop the general G path. But for that sliver of time, the force direction has changed, even if it has not crossed over the zero point, and so must the onset cue, depending on softness chosen. The data graph is only showing what direction the momentary G-force is heading. It is this data that is being fed to the simulator and hence, it too should move appropriately.

Now for sure the driver in the sim would not want to feel every minute motion, therefore a certain amount of response softness is desired. But that is totally dependent on the driver, at least from all my experiences.

These fluctuations are so close in time that it may seem like one sustained motion, but it is not. Try this with your road car; Driving smoothly around a curve, jerk the steering this way and that. You will feel the G effects by being tossed about, but you are still heading around the same curve. You have not necessarily crossed over the zero point. And that is why a washout theory works, just that it involves larger time spans and rates of changing directional forces.

As much as drivers would like their race cars to feel like it is being driven on the proverbial rail, it does not happen when on the edge of adhesion. And if it is happening, then you are driving too slow!

So the jury is still out on my rhetorical question, how little time can the brain be fooled into sustained motion, and how much is truly needed?

R-eng

I think we're asking the same question in different words. I am looking into the concept of extending a cue and you are refering to fooling the brain into sustained motion.

For me, this is complicated by the fact that I have to determine the initial cue as well as determine an appropriate mix of muscle and skeletal cues with these motion cues. I am confident that the proper combination of muscle / skeletal cues mixed with motion cues will provide better immersion than either could achieve alone. I think this fundemental difference will make it difficult to compare notes.

One of the mysteries that remains for me is how to best allow personalization since we've already estabished that these things vary by person.

I forsee some kind of simulator setup wizard in an upcoming version of the Commander software that walks the user through a series of questions and gathers their feedback about a series of visuals and movements. The result of this would be used to offset a consistently produced set of motion profiles to that users preference. The concept is simple enough, but at the moment, the correct formula eludes me and the amount of user feedback required to pull it off is overwhelming. Nontheless, I feel it is a worthwhile endeavor that would help to keep user perception of X-Sim's motion quality a step above the competition.

As for the G-line when on the edge of traction, I generally represent most of this in the rear traction loss axis and only a small amount in the seat.

BTW- what sim design are you using? I don't recall ever seeing it.