Date: Thu, 19 Sep 2002 13:49:37 -0500 From: "Rob King" Subject: Re: 0GNav Sender: To: "Chad Wingrave" Cc: "3D UI list" <3d-ui@hitl.washington.edu> Message-id: <3D8A1C41.7729A7F7@bellatlantic.net> MIME-version: 1.0 X-Mailer: Mozilla 4.78 [en] (Windows NT 5.0; U) Content-type: text/plain; charset=iso-8859-1 Content-transfer-encoding: 7BIT Importance: Normal X-MSMail-Priority: Normal X-Accept-Language: en X-MimeOLE: Produced By Microsoft MimeOLE V4.72.3110.3 X-Authentication-Info: Submitted using SMTP AUTH PLAIN at pop015.verizon.netfrom [128.82.86.180] using ID at Thu,19 Sep 2002 13:49:29 -0500 References: X-Authentication-warning: torch.hitl.washington.edu: majordom set sender toowner-3dui@hitl.washington.edu using -f X-Priority: 3 (Normal) Chad, As you point out, there are a *lot* of issues. Chad Wingrave wrote: > > Traditional navigation techniques might work with a few twists to account > for the extra DOF that you need. You can always twist a tracker or change > its pitch, most navigation techniques ignore these DOF and should map > naturally (just be careful to have dead-space in the rotations). If you > have experts in the system, two-handed might work well as they can control > more DOFs at once. If you have users you want to impress, your > wall-pushing might work as they might feel more like they are floating in > space (if that is your desired effect) but if you are not trying to get a > training transfer or impress, no reason dealing with the limitations of > reality. Why walk when you can fly? > You are correct. There are one- and two-handed techniques that can be applied. One key issue is the number of DOF to control with a single gesture. From experience, coupling too many DOF to a one-handed control can be problematic. Stability and joint limits are among the issues here. On the other hand, a 'rope-pull' or 'fly me in this direction' sort of navigation metaphor might compensate by being more natural/intuitive. Two-handed controls require separation of input. One might control orientation with one hand (non-dominant better?) and velocity or position with the other. I'm not sure if separation of input means greater or less cognitive load on the user. A related issue is whether to use fixed devices (e.g. joysticks attached to the chair) or free-space devices (e.g. wanda) for interaction. So, what do you think? One- or two-handed? Free-space or fixed devices? > Since you are in simple spaces (depending on which design of the space > station you use), the following might not be a problem but if the spaces > are large.... > I would imagine your biggest problem will be keeping the user aware of > where they are. After a few minutes in a sizable area, they will most > likely get lost and their biggest problem of navigation will not be > travel but wayfinding. You might want to consider adding in feedback such > as a small alphed-out WIM at the screen bottom, directional pointers to > user-specified waypoints (like bread crumbs) or something similar. > A navigation aid is a good idea, but I'm pretty sure that it is ruled out by the experimental design. Recall that the objective is to measure physiological responses to the task of orienting oneself in 3D. The navigation aid would provide an artificial reference frame... Wayfinding is also a concern, but we do not want it to be an issue. It shouldn't be one in the first experiment. In the second, we will need to be careful that subjects are thoroughly familiarized with the station layout by training sessions. > At the cognitive level, I don't think people's brains have any practice at > all in the new dimensions you are adding so assisting that might be your > biggest hurdle. Might want to check out what spelunkers, cave divers, > astronauts or gamers of Descent do to help them get through their > environments. > Descent was one of the environments I was thinking of! (My son regularly kicks my butt--a killer interface that would even my chances against him is, of itself, a worthy project.) On the psychological front, there are a host of issues in sensation, perception and cognition that are at the heart of this. For example, it has been shown that humans are by and largely incapable of mentally conceptualizing the optimal path that rotates an object from one orientation to another, unless that path coincides with a natural axis. A natural axis is one that is a) vertical, b) horizontal, or c) in the direction that the user is looking. The consequence is that users tend to use a sequence of natural axis rotations (sort-of like thinking in Euler angles instead of quaternions). However, it is unclear whether those findings translate to the task of orienting oneself vice orienting an object. So, would you expect subjects to be able to visualize the optimal path? Rob > -Chad Wingrave (cwingrav@cc.gatech.edu) > Georgia Tech, Graduate Student > http://www.cc.gatech.edu/~cwingrav > > On Wed, 18 Sep 2002, Rob King wrote: > > > The subject line should be read, "zero Gee nav", as in navigation in a > > zero gravity virtual environment. I'd like to hear your thoughts on the > > navigation techniques that might be employed. > > > > We are in the initial planning stage of software development for a > > project that involves navigating and interacting in an immersive > > environment. We want to collect data concerning physiological responses > > to the task of orienting oneself spatially in 3 dimensions, and then > > navigating from one location to another. We must choose the interaction > > devices and > > techniques that will be used. We can assume we have some sort of button > > device with 6 DOF tracking. > > > > Two experiments are envisioned. The first involves orientation and > > simple navigation in a virtual room where each wall is textured with a > > large checkerboard pattern. The room is rectangular and the walls are > > different colors. Subjects will be initialized at an arbitrary position > > and orientation. The task will be to navigate to a new position. > > > > The second involves orientation, navigation and interaction with objects > > in a model of the international space station. Interaction is simple, > > consisting of 'touching' one or more wall switches in the virtual > > environment. > > > > Training transfer is not an important issue, but we'd like to implement > > reasonable techniques. For example, we've all seen pictures of > > astronauts pushing off one wall and arresting motion on another. A wall > > touch technique based on collision detection comes to mind, but the > > control and feedback issues are complex. > > > > We've got several other ideas, but I'd like to hear your unbiased > > thoughts on the techniques that could be applied. > > > > Thanks 1.0E+06,