LHETTINGER@FALCON.AAMRL.WPAFB.AF.MIL (01/18/91)
Among the many benefits to be derived from developments in virtual reality there is at least one potential negative side effect that needs to be addressed - the problem of motion sickness. For the last several years a number of us, working primarily under the sponsorship of the Army and Navy, have been investigating the phenomenon of motion sickness in flight simulators. We have been particularly interested in trying to determine the causes of sickness in situations where the is no physical motion of the user, i.e., where all the motion is visually specified. We have recently published data indicating that only those people who experience strong illusory sensations of self motion (traditionally refered to as "vection" in the psychophysical literature) will experience problems with sickness. The vection illusion consists of the overwhelming sensation that one is moving despite explicit knowledge to the contrary. The quality of the illusion of motion that is imparted to the user is often taken as an informal "measure" of the quality of a simulator and, presumably, other types of virtual imaging devices whose design intentions are similar. Our concern centers primarily around two issues: (1) the quality of training afforded by simulators that make a significant number of people sick, and (2) safety concerns for users hours after leaving the virtual environment. While data on the long-term effects of "simulator sickness" is scarce, there are numerous anecdotal accounts of these problems. One of the more dramatic concerns an Air Force pilot who, upon driving home after an extended training session in a flight simulator, experienced a 180- degree inversion of the visual field. Fortunately, he was able to safely stop the car and recovered soon afterward. Questions of user well-being, to say nothing of potential liability consideration, make this a rather critical issue for the development of virtual reality. I am anxious to communicate, share literature, debate issues, examine solutions, etc. with any and all interested individuals. Larry Hettinger Logicon Technical Services Inc. (513) 255-8770 P.O. Box 317258 Dayton, Ohio 45431-7258 lhettinger@falcon.aamrl.wpafb.af.mil
sharp@fsd.cpsc.ucalgary.ca (Maurice Sharp) (01/19/91)
This is not news. Meridith Bricken at the First Conference on Cyberspace made the same comments. Just because you are in a virtual world, does not mean you can get rid of cues we use. That means you need a floor, an up, a down, etc. It does not mean that future generations will need these things, but we certainly do. maurice -- Maurice Sharp MSc. Student (403) 220 7690 University of Calgary Computer Science Department 2500 University Drive N.W. sharp@cpsc.UCalgary.CA Calgary, Alberta, T2N 1N4 GEnie M.SHARP5
zweig@cs.uiuc.edu (Johnny Zweig) (01/19/91)
In a psych class I took a few years back, we saw a film showing people placed in the center of an almost spherical room with a random blotchy pattern displayed on the walls/floor (picture a guy standing on the inside bottom of a 20' beachball). The pattern was slowly rotated in a manner that simulated how things would look if the subject were falling over, though of course the floor stayed steady as a rock (or floor, as the case may be). It was fascinating to watch the subjects as they tried to just stand there despite their eyes telling them they were falling over. Most of them ultimately flung themselves to the side trying to correct for the nonexistent fall, even though the floor was padded and they were told the room was not moving. Seen from the camera, it is amazing to watch the people shuddering, trying not to move, then leaping viciously at the floor to "keep from falling". The relation between the visual sensorium and the sensation of motion is really amazing. -Johnny Vision
jwtlai@watcgl.waterloo.edu (Jim W Lai) (01/19/91)
In article <1991Jan18.185231.17365@cpsc.ucalgary.ca> sharp@fsd.cpsc.ucalgary.ca (Maurice Sharp) writes: > This is not news. Meridith Bricken at the First Conference on >Cyberspace made the same comments. Just because you are in a virtual >world, does not mean you can get rid of cues we use. That means you >need a floor, an up, a down, etc. > > It does not mean that future generations will need these things, >but we certainly do. Ah, but when are these cues learned? And what effect would not learning these cues we take for granted have when dealing with material reality?
tomh@cs.fau.edu (Tom Holroyd) (01/23/91)
NASA has developed drugs for combating space sickness. Some of these might be useful for VR motion sickness. Tom Holroyd tomh@bambi.ccs.fau.edu
LHETTINGER@FALCON.AAMRL.WPAFB.AF.MIL (01/24/91)
>NASA has developed drugs for combating space sickness. Some of >these might be useful for VR motion sickness. NASA has also had some success with Biofeedback techniques - Dr. Patricia Cowings at NASA Ames has done the majority of that work. I don't think either of these approaches are going to work particularly well for individuals whose use of simulators or other VR devices, either for work or entertainment, is fairly limited. Biofeedback is time-consuming and may deteriorate with lack of exposure to the offending environment, while anti-motion sickness medications often have deleterious side effects (eg., drowsiness). The solution to the problem will probably have to come from changes in the design of visual displays that appear to cause the problem, or in approaches to how the systems are used. As an example of the latter approach, it is often best to acclimate a pilot to a new simulator before "turning him loose" to do whatever he/she wants in the thing. During the initial phases of exposure to a virtual motion environment, it appears to be best to gradually adapt to the new situation. For a pilot, this may mean doing nothing more intense then flying straight and level for a while on the first day or two, and then gradually increasing the intensity of maneuvering as adaptation sets in. Limited exposure times also appear to be critical for avoiding sickness in novice users. Larry Hettinger
LHETTINGER@FALCON.AAMRL.WPAFB.AF.MIL (01/24/91)
It seems to me that there are four factors which have the capacity for causing motion sickness in V.R.: 1) Difference between perceived gravity and viewed "down." 2) Difference between perceived angular velocity and acceleration and viewed angular velocity and acceleration. 3) Perceived linear velocity outside of normal experience. 4) Difference between perceived linear acceleration and viewed linear acceleration. 5) Changes of direction at high speed. 1 and 2 are very easily solved. Just make the viewed orientation an accurate representation of the actual orientation. To turn, you actually have to turn your head; to tilt you actually have to tilt your head; the ground plane that you see remains true. I think you could probably get away with a lot here, such as putting a scaling factor between your real head's yaw velocity and your virtual head's yaw velocity. We are relatively desensitized to 3 when we are moving in the forward direction due to our experience driving or riding in cars. Forward motion is probably the easiest to which to become desensitized, as it carries no message that can be interpreted as "rotation." People who ride in trains a lot are probably desensitized to other directions of motion as well. However, I think it's probably a good idea to limit purely virtual motion to forward, for user interface considerations as well as perceptual ones. The concepts of walking and running are more easily understood than the cancrazan. 4 is tricky, but it may not be that much of a problem. Acceleration only need last until you get up to speed. Perhaps the disorientation can be minimized by choosing an acceleration so that getting up to speed is neither too abrupt nor too prolonged. If this is not sufficient, either viewed acceleration can be mapped through a scaling factor to actual acceleration (i.e. you have to walk forward to start accelerating, and stopping walking will only slow you down a little bit.), or acceleration can be triggered by tilting back the head, which emulates some of the sensations of acceleration. We might also be able to eliminate the need for such high perceptual speeds by compressing intervening space--do you really need to see every grain of sand in the Arizona desert when zipping from Phoenix to Tucson? I would personally like a set of virtual Seven League Boots. I don't know about 5, but I think that, as abrupt changes in direction at higher than brachiating speeds haven't yet contributed to much evolutionary pressure, this can maybe be ignored. If not, one could constrain the direction change to be gradual enough not to matter and require a slowdown before speedier direction changes. This would provide a nice kinesthetic parallel to driving an automobile. I think I'll go down to Disneyworld next weekend and do some research on Star Tours and the World of Motion, not to mention the Interstate. :-) Eric Pepke INTERNET: pepke@gw.scri.fsu.edu Supercomputer Computations Research Institute MFENET: pepke@fsu Florida State University SPAN: scri::pepke Tallahassee, FL 32306-4052 BITNET: pepke@fsu Disclaimer: My employers seldom even LISTEN to my opinions. Meta-disclaimer: Any society that needs disclaimers has too many lawyers.
LHETTINGER@FALCON.AAMRL.WPAFB.AF.MIL (01/24/91)
Eric Pepke recently wrote: >It seems to me that there are four factors which have the capacity for >causing motion sickness in V.R.: > 1) Difference between perceived gravity and viewed "down." > 2) Difference between perceived angular velocity and acceleration > and viewed angular velocity and acceleration. > 3) Perceived linear velocity outside of normal experience. > 4) Difference between perceived linear acceleration and viewed linear > acceleration. > 5) Changes of direction at high speed. I agree. All of the things you mention involve visual-vestibular conflicts which appear to be present in most motion-sickness situations, not just flight simulator sickness. I would also suggest a sixth potential cause: head movements. Any motion of the head made in the presence of conflicting visual information for self motion (i.e., the eyes "specify" rotational motion, the inner ears "specify" some different motion) may serve to enhance the conflict. Whatever the cause - excessive head movements often lead to trouble with sickness. >1 and 2 are very easily solved. Just make the viewed orientation an accurate >representation of the actual orientation. To turn, you actually have to >turn your head; to tilt you actually have to tilt your head; the ground plane >that you see remains true. I think you could probably get away with a lot >here, such as putting a scaling factor between your real head's yaw velocity >and your virtual head's yaw velocity. I think this is a reasonable suggestion for some VR settings - for flight simulators, though, I don't think it will work. To link the simulator dynamics to head dynamics would violate the training purposes of the device. >We are relatively desensitized to 3 when we are moving in the forward direction >due to our experience driving or riding in cars. Forward motion is probably >the easiest to which to become desensitized, as it carries no message that >can be interpreted as "rotation." People who ride in trains a lot are probably >desensitized to other directions of motion as well. However, I think it's >probably a good idea to limit purely virtual motion to forward, for user >interface considerations as well as perceptual ones. The concepts of walking >and running are more easily understood than the cancrazan. It is interesting to note that experienced pilots appear to have much more difficulty with simulator sickness than inexperienced pilots. This finding appears to support what you are saying above and I think can be interpreted in the following way. Pilots become very sensitive to certain visual-vestibular consequences of controlling their self motion, and therefore can be expected to be very sensitive to violations in these relationships. Specifically, when a pilot pulls into a turn in flight it produces a characteristic pattern of visual-vestibular stimulation which is very specific to that situation. Through perceptual learning, the pilot becomes highly attuned to that pattern of stimulation. We can replicate the visual part of that pretty well in a simulator, but we can't approach the vestibular part. And therein lies the problem.... >4 is tricky, but it may not be that much of a problem. Acceleration only >need last until you get up to speed. Perhaps the disorientation can be >minimized by choosing an acceleration so that getting up to speed is neither >too abrupt nor too prolonged. If this is not sufficient, either viewed >acceleration can be mapped through a scaling factor to actual acceleration >(i.e. you have to walk forward to start accelerating, and stopping walking will >only slow you down a little bit.), or acceleration can be triggered by tilting >back the head, which emulates some of the sensations of acceleration. We >might also be able to eliminate the need for such high perceptual speeds by >compressing intervening space--do you really need to see every grain of sand >in the Arizona desert when zipping from Phoenix to Tucson? I would personally >like a set of virtual Seven League Boots. Another method of dealing with this may be to provide a brief inertial input to the user with a dynamic seat of some sort. This would provide them with a transient vestibular input that would mimic the short time constant of the vestibular system in linear acceleration. Then you can let the visual take over. >I don't know about 5, but I think that, as abrupt changes in direction at >higher than brachiating speeds haven't yet contributed to much evolutionary >pressure, this can maybe be ignored. If not, one could constrain the >direction change to be gradual enough not to matter and require a slowdown >before speedier direction changes. This would provide a nice kinesthetic >parallel to driving an automobile. Well, I don't think you want to ignore this because it's still a visual-vestibular conflict situation and therefore can't be ruled out as potentially contributing to disorientation and sickness. This may also be a potential use for a dynamic seat of some sort. >I think I'll go down to Disneyworld next weekend and do some research on >Star Tours and the World of Motion, not to mention the Interstate. :-) Just don't move your head around too much 8-) Larry Hettinger
pepke@ds1.scri.fsu.edu (Eric Pepke) (01/24/91)
In article <14998@milton.u.washington.edu> LHETTINGER@FALCON.AAMRL.WPAFB.AF.MIL writes: >flight simulator sickness. I would also suggest a sixth potential cause: >head movements. Any motion of the head made in the presence of conflicting >visual information for self motion (i.e., the eyes "specify" rotational >motion, the inner ears "specify" some different motion) may serve to enhance >the conflict. Whatever the cause - excessive head movements often lead to >trouble with sickness. I thought what I said included this. The semicircular canals are definitely part of the nonvisual perception that needs to be squared with visions. Excessive head movements can cause another thing entirely--they can make the fluid slosh around in such a way that it has no longer anything to with reality and cannot be predicted. (Hmmm... I wonder what would happen if one were to model the head tilt sensors after semicircular canals?) >>1 and 2 are very easily solved. Just make the viewed orientation an accurate >>representation of the actual orientation. > >I think this is a reasonable suggestion for some VR settings - for flight >simulators, though, I don't think it will work. To link the simulator >dynamics to head dynamics would violate the training purposes of the device. You're right. I am specifically thinking about forms of virtual reality where emulation of an actual physical environment is not important. I'm thinking of sort of an idealized reality, for the same reason that I don't think my Macintosh "desktop" needs to have coffee stains. Personally, I think that if you're going to have a flight simulator, you should at least spring for one of those virtual roller coasters that run on hydraulics (the cheaper ones on motors and worm gears) one sees in all the malls. It'll set you back a lot less than what the VPL guys will soak you for. >Well, I don't think you want to ignore this because it's still a >visual-vestibular conflict situation and therefore can't be ruled out as >potentially contributing to disorientation and sickness. This may also be a >potential use for a dynamic seat of some sort. I just thought of another potential solution (again, only in my kind of VR). The only reason that we need to make really sharp turns in the world is that things are fixed in their coordinate system. Why not, when I turn, warp the positions of distant objects so that a perceptually small turn over a large distance generates a big turn? Incidentally, a while back for my day job I implemented a way of flying around a dataset in a visualization package I'm working on. I wound up emulating the dynamics of some sort of idealized airplane because it seemed to feel the most natural. I'm not sure why, as I've only flown an airplane a couple of times for short periods, but maybe it's because the dynamics are similar to riding a bicycle. The more I think about, the more I want to approach reality like a salad bar-- take what I like and leave the rest. My VR house can look like a physical entity, but why can't it be on Pepke block of Name street and Eric block as well? Why shouldn't it be on the intersection of an arbitrarily large number of streets, name street and occupation street and hat size street? Why shouldn't it have an address in Jazztown and Nerdville and Punk City and the red light district as well? Why shouldn't it be tardis-like on the inside? The trick and the challenge is to abstract the feelings of reality in such a way that they still work but don't have the same constraints. Eric Pepke INTERNET: pepke@gw.scri.fsu.edu Supercomputer Computations Research Institute MFENET: pepke@fsu Florida State University SPAN: scri::pepke Tallahassee, FL 32306-4052 BITNET: pepke@fsu Disclaimer: My employers seldom even LISTEN to my opinions. Meta-disclaimer: Any society that needs disclaimers has too many lawyers.