Biomedical Engineering Reference
In-Depth Information
Fig. 16.9
The process we
will be going through to
translate from the data a
controller provides to a virtual
locomotion solution
16.3.1 Understanding the Data Coming from the Device
The first step in working with a new device is accessing its data and understanding
what the reported values mean. Each device will have a different software mechanism
for obtaining the data, specific to that device, but once you obtain your data stream,
you need to understand what it means, its limitations and decide how best to make use
of it. The game controller discussions above demonstrate how their creators cleverly
synthesize useful data for a developer. In many cases, such as with the Wiimote, you
may never have access to how the device actually does this, only having the raw data
as input. Or, you may be able to find a toolkit that does much of this synthesis for you
and it may be sufficient for your needs. In the case of the Kinect, Microsoft provides
a software development kit (SDK) to provide the raw depth and color image as well
as skeletal data interpreted by Microsoft's own algorithms. In this situation, you can
choose whether to perform an analysis on the raw images or use the skeletal data for
gesture recognition. Choosing to performyour own analysis should not be undertaken
lightly, as it is complex and time consuming. However, many good reasons exist for
this such as needing a classifier for a specific gesture, needing a new feature in the
data stream recognized or needing functionality in the current toolkit that does not
exist. In these cases, you will need to dig deeper into the data and understand it. Once
you know the data, you can make a better choice in the algorithm to use. This is the
topic of the next section, where we show how some algorithms work better with raw
data while others may rely on a higher-level interpretation to already be known.
