Week 4, Part 2: 9 - 16 March [Understanding Kinect practical capabilities]

-

Using the prebuilt applications included in the development kit v1.8 I could test the accuracy of the Kinect.

While both reasonably reliable, facial tracking and voice input will be impractical for this project. Facial tracking is very restricted in terms of suitable content, and with the potential for many people to attend the event, there is a high chance voice input will be hindered by nearby speech.


Head tracking and facial recognition
Voice input

While overall the device is responsive, fast movements often cause the Kinect to lose joints, so the user must stand in the calibration pose until redetected.

If the Kinect is positioned poorly, with obstacles blocking its view of the user, when the majority of a limb isn’t tracked it makes detection of the remainder of joints unreliable and often requires recalibration.

There are some gestures where limbs pass behind/in front of each other that if executed too slowly achieve blocking like the above, occasionally resulting in the same issue, whereby failed tracking of joints can produce unintended results – although if tracking is lost briefly in this way, recalibration is rarely required with full skeletal tracking being accurately recovered within seconds of all joints becoming uninterrupted. I will consider the necessary gestures the user will need to use as input to avoid this as much as possible.

Lost limb tracking
False detection with single user
Another issue where occasionally the sensor will detect additional false sets of joints, when there is only one user (and no other person in the vicinity). I found this occurred very infrequently, but should it happen during the exhibition application, theres potential for it to interfere with the user experience.

The sensors can be calibrated for users in a seated or standing stance, which may be of help when it comes to the exhibition as we are likely to have potential users of all ages. Though if the user is seated in a high-back chair the sensor can often struggle to clearly differentiate between the user and the chair, again requiring the user to adopt the calibration pose.

The sensor has a maximum range, but also found that if users are too close to the Kinect tracking is lost. The documentation confirmed that the Kinect v1 has an optimal tracking area of 6m2 in front of the Kinect, though this begins from approximately 90cm from the sensor. The positioning of the sensor will be something to consider when creating applications for the exhibition.

Gesture input
Skeletal tracking

Depth tracking










The most appropriate applications will focus on skeletal tracking (primarily gestures) and depth tracking (point cloud). Both offer a range of potential implementations, suitable for the exhibition.

Comments

Post a Comment

Popular posts from this blog

Week 3: 2 - 9 March [Continued AR prototype]

-
Image
I have spoken with the university’s 3D technologist, who is happy to assist with printing the airship model, but does not have the capacity to do so for several weeks. Until I can print a model I will continue to develop the prototype started last week. Instantiating model versus instantiating model with shadow shader. To add an element of realism to the instantiated airship, I added a plane below the model so the light source within the scene would be blocked by the model, casting a shadow onto the plane. The shadow had very poor definition, not appearing to look even vaguely similar the model, and often went unnoticed against grey/white backgrounds. To improve this, I added a shader to the airship model, which casts a much more well-defined shadow. I also updated the components of the plane within Unity, setting its alpha value to 0, making it transparent. This still casts a shadow onto the planes surface, while making it appear to the user that the shadow is cast direc...
Read more

Week 7, Part 2: 30 March - 6 April [Leap Motion, Prototype]

-
Image
Thinking of possible uses of the Leap Motion in the application, I ruled out a pop-up menu as this could only be used to trigger scene changes – not the level of engagement the project requires. Despite this, I was still able to draw inspiration from the ‘Hover-UI-Kit’ menu I experimented with in the previous blog entry, in Unity. Using a gesture to activate a ‘mode’ then more intuitive movements within the mode for further functionality. Using the hand to select parts of the airship to then display related information was the idea I decided to develop. This could be done with relative ease through raycasting out from the hand prefab to detect colliders on a ship model. I used the last bone in the index finger for the origin of the ray, casting the ray forward in the direction of the bones Z-axis. Once this functionality was implemented it needed to be made toggleable, triggered by a user gesture. Leap Motion provides an example script to detect predefined hand positions. Th...
Read more

Week 1: 16 - 23 Feb. [Meeting with Museum Director]

-
Image
While I was intrigued by both the energy efficient car and the museums interactive experience, the prospect of working with new technology directly related to the content of the degree, where people outside of the university would be the end users made the museum’s project far more appealing. Diss museum had reached out to the university for assistance with their upcoming exhibition, to commemorate the successful double Atlantic crossing of the R34 airship. No brief or requirements were specified in the museum’s request, so after learning of the project I contacted the museum manager, Basil Abbott, to arrange a meeting to further clarify what they hoped to include in their exhibition and available resources. From initial email exchanges it was clear that the intention was to involve the community with a hands-on experience(s) rather than a static exhibition, the implementation of which the museum has left to our discretion. To prepare for the meeting it was necess...
Read more