My thesis work involved me identifying the trade-off of usability to fidelity in a modern VR set-up, in the domain of user training.
In order to study the trade-offs between usability versus fidelity in a VR system, I had to develop a novel VR system. Using the Oculus Rift (DK2), for visual fidelity and a depth camera (Creative DepthSense) for finger tracking, I constructed a system where the user can naturally interact with objects virtually.
Challenges:
- Finger-tracking : This took a major part of my development time, getting the perfect position and filtering the tracking information for reliable and consistent tracking.
- Interaction : I needed to develop an intuitive and easy to understand method of picking-up and dropping Lego. Since the study is about training, the interaction technique must be as natural as possible and should not pose an obstacle for the user in terms of interaction fidelity, thus there needed to be an algorithms which interprets the actions of a user based of their tracking information.
- Calibration system : The user would be interacting with a 3D world when they are in the space where they will be doing the real world tasks, thus the tracking of the hands should be relatively in the same position as they would expect their hands would be in the real world. Hence we needed a calibration system which ensured that unexpected offsets did not exist when the user uses the system.
- Real world modeling : I am not a 3D artist and have not taken any course in it. I self taught myself to model relatively good game assets in Blender. Using these skills I needed to model the 3D world accurately and to scale, and also get the texturing and shading right (my lighting was a bit off). The scale of the objects was the most important attribute to get right. Since the virtual world, should be identical to the real world environment, the scale of the system needed to be just right.
- Correctness and monitoring algorithm : I needed to develop a system which would record all the events in the system to a flat file which I could then filter and analyze in excel. More over, I also needed a sort of correctness system, which would identify if the user has placed the pieces in the right spot or not.
Bear in mind that this is a work in progress, there are a lot more assets to model bugs to fix.
Software packages used:
- Unity 3D
- Intel Perceptual Computing SDK 2013
- Blender
- Photoshop
Being a solo project I had to do a lot of research and a lot of prototyping. Trail and error tests, for which we finally came up with this solution. What you see is around 30% of the total work I had done over the span of my thesis. This was my first step into the world of Research and Development, and I must say have thoroughly enjoyed every minute of it, at times it was frustrating but in the end very satisfying.
In order to study the trade-offs between usability versus fidelity in a VR system, I had to develop a novel VR system. Using the Oculus Rift (DK2), for visual fidelity and a depth camera (Creative DepthSense) for finger tracking, I constructed a system where the user can naturally interact with objects virtually.
Challenges:
- Finger-tracking : This took a major part of my development time, getting the perfect position and filtering the tracking information for reliable and consistent tracking.
- Interaction : I needed to develop an intuitive and easy to understand method of picking-up and dropping Lego. Since the study is about training, the interaction technique must be as natural as possible and should not pose an obstacle for the user in terms of interaction fidelity, thus there needed to be an algorithms which interprets the actions of a user based of their tracking information.
- Calibration system : The user would be interacting with a 3D world when they are in the space where they will be doing the real world tasks, thus the tracking of the hands should be relatively in the same position as they would expect their hands would be in the real world. Hence we needed a calibration system which ensured that unexpected offsets did not exist when the user uses the system.
- Real world modeling : I am not a 3D artist and have not taken any course in it. I self taught myself to model relatively good game assets in Blender. Using these skills I needed to model the 3D world accurately and to scale, and also get the texturing and shading right (my lighting was a bit off). The scale of the objects was the most important attribute to get right. Since the virtual world, should be identical to the real world environment, the scale of the system needed to be just right.
- Correctness and monitoring algorithm : I needed to develop a system which would record all the events in the system to a flat file which I could then filter and analyze in excel. More over, I also needed a sort of correctness system, which would identify if the user has placed the pieces in the right spot or not.
Bear in mind that this is a work in progress, there are a lot more assets to model bugs to fix.
Software packages used:
- Unity 3D
- Intel Perceptual Computing SDK 2013
- Blender
- Photoshop
Being a solo project I had to do a lot of research and a lot of prototyping. Trail and error tests, for which we finally came up with this solution. What you see is around 30% of the total work I had done over the span of my thesis. This was my first step into the world of Research and Development, and I must say have thoroughly enjoyed every minute of it, at times it was frustrating but in the end very satisfying.
Comments
Post a Comment