DaBirkmeier

Chair for Computer Aided Medical Procedures & Augmented Reality
Lehrstuhl für Informatikanwendungen in der Medizin & Augmented Reality

THIS WEBPAGE IS DEPRECATED - please visit our new website

Implicit guidance to poses in AR - Implizite Führung zu Posen in der AR.

Motivation

"Real-world wayfinding has been researchend extensively, with studies of aids like maps, directional signs, landmarks, and so on. In virtual worlds, wayfinding can also be crucial in a large complex environment, an efficient travel technique is of no use if the traveler has no idea where to go." - D. BOWMAN

With the possibilities in Augmented and Virtual Reality people need more than guiding in the huge and growing space of the new applications. Additional to achieve the right position the user must also look in the right direction.

Abstract

Target of the system is guidance to a pose. You're not only guided to a position the field of view is also guided to a defined orientation. The guiding uses only five dimensions, three for the position and two for the orientation. The sixth dimension would be the rotation of the head along the viewing direction, for the users' convenience it is free of guiding.

Also the guiding should happen implicit with minimal cognitive effort.

To achieve guiding in 5 dimensions the form of the path must be rotationally-symmetrical.

HMDfront

Visualization

The basic idea is the projection of a path. Using a 3D optical device the path fits into environment and can show you your way. By simply following the path you accommodate to the given pose. This is happening implicit without additional hints and clues.

The following pictures are take through the head mounted display and showing the final version of the path.

HMDseethrough1 HMDseethrough2 HMDseethrough2

The appearance of the path was created and checked in a formative test with several users with and without expert knowledge in Augmented Reality. One of the biggest issues was the mounting point, the starting point of the path, and the height of the path.

For the mounting point several possibilities were tested, mounting at floor level, like a rail, to mounting at ceiling level and everything in between. To prevent the user from getting lost by losing the sight to the path a mounting point in the field of view is necessary.

Chestmounted Foreheadmounted Nosemounted

The users liked most, when the path was close to the tip of their nose and their mouth. If the path starts so close to the face it must be small to avoid blindfolding the eyes.

In natural forward movement people look slightly to the ground. This defines the height of the path some centimeters under the height of the users' eyes. Therefore the height of the path is calibrated for every user.

User Test

The System was evaluated with a user test. To check the possibilities of the system three test scenarios were build. Checked were the aspects of navigation, directing the field of view and the guiding to poses. Because of the small tracking volume of the laboratory these scenarios work within a very small space.

Search Scenario

The search scenario focused on guiding to poses. Only at the correct position with the correct orientation you could find something hidden in the lab. To have more hiding-places a box-tower was placed in the lab. In two different tasks the users had to search guided and unguided for a virtual ufo. Within the unguided search the hideout places of the ufos were easier than within the guided search.

guided1 guided2 guided3

unguided1 unguided2 unguided3

This video shows the guided search and should explain the user test.

Photography Scenario

The photography scenario focused on the field of view of the user. The users had to take pictures of a virtual Ufo with three different tasks. Two Tasks guided the user to three different poses in which the user had to take the pictures. The third task was without guiding, the task was to "take 3 pictures in such a way, so anyone can rebuild the ufo without having seen it."

The two sets of guided photos were necessary to check for learning and copying effects of the guided poses in the unguided task. For more details feel free to read the thesis :-).

The following pictures show the guiding to a photo pose.

Photo1 Photo2

This pictures show the result of the photo task, the ideal photo, a bad and a good example of user pictures.

1_3ideal 1_3bad 1_3good

The left half of the pictures shows the left image of the HMD, the right half the right image. The Ufo is centered in the left image und partial cut in the right image, this could also happen the other way. This effect is created by the dominant eye of the user. These pictures are taken by people having a dominant left eye, the right eye is supporting the left eye. Therefore the picture of the left eye shows the Ufo centered.

This video shows how the test worked.

Maze Scenario

The maze scenario focused on the navigational guidance of the path. Because of the small tracking volume a pillar maze was build with the path winding through the pillars. To check the navigational behavior in three dimensions the path changed his height level so the user had to bend and stretch while following the bath.

Maze Maze1 Maze2

This video should give a good impression how the scenario worked.

Additional obstacles were implemented to support the guidance provided by the path. This video should show the difference

Summary

The results of the user tests are discussed in detail within my work. Diploma Thesis Martin Birkmeier

Additional Stuff

Some additional thinks I wanted to mention.

Hardware

The HMD used was a nVisor ST60 Link to NVis.

The tracking was done by a separate PC using DTrack. Therefore a 3D marker was mounted at the HMD, this marker was tracked with four cameras mounted at the ceiling and the side of the lab.

HMDfront HMD

Calibration

In normal use the HMD has to be calibrated for every user. Without knowledge how this calibration works it can take a while and you can't verify the user is seeing what you want him to see. Therefore a dummy configuration for the System was done. But with the dummy configuration and a lot of possibilities to wear the HMD in a way that you can't see properly through the lenses a check was necessary to control if the user sees what he is supposed to see.

A virtual table was built matching a real table. This table has red markers matching invisible real markers on the real table. The user is given the task to tab the virtual markers, if he tabs the real markers on the real table the HMD is worn in the right way.

HMDfront HMDfront HMDfront

Stereo Eyesight Test

To make sure every user has the same chances we had to make sure the every user can see in stereo. Besides the obvious problem, being one-eyed or crossed-eyed there is the disability to see in stereo. The brain isn't able to create a 3D image of the two images the eyes deliver. The percentage of affected people many numbers are named by different sources, most of them name 5-10%.

There are some tests for Stereo Eyesight, like the Lang- or Titmustest. These tests were too expensive to acquire for this work, so with the basic idea of these tests I created my own Sight Test.

On several pages you see 3D-anaglyph images mixed with fake 3D images. If someone can name the real 3D pictures over all test pages without any errors his ability for stereo sight is secured. For this test you have to wear green-red anaglyph glasses. The test can be found in the Download section.

langtest

Download

Download Diploma Thesis Martin Birkmeier

Download Stereo Sight Test

Download all the pictures on this site in original resolution. Only use them with proper reference to this work

For any kind of questions please contact me Martin Birkmeier.

ProjectForm
Title: Implicit Guidance to Poses in Augmented Reality
Abstract: Developing a guidance system for a see-through stereo-view head mounted display. The system should guide to a position and a pose. A pose is specified through the orientation and the facing of the user. The System projects a path from the user's position to the target into the reality. By following this path the user is implicit lead to a specific pose without any extra effort.
Student: Martin Birkmeier
Director: Gudrun Klinker
Supervisor: Marcus Tönnis
Type: Diploma Thesis
Area:  
Status: finished
Start: 2010/08/15
Finish: 2011/02/15
Thesis (optional): http://campar.in.tum.de/twiki/pub/Students/DaBirkmeier/da-birkmeier.pdf
Picture:  


Edit | Attach | Refresh | Diffs | More | Revision r1.32 - 15 May 2011 - 19:34 - MartinBirkmeier