1This directory contains a simple python script for visualizing
2the behavior of the WindowOrientationListener.
81. Python 2.6
92. numpy
103. matplotlib
12eg. sudo apt-get install python-numpy python-matplotlib
18The tool works by scaping the debug log output from WindowOrientationListener
19for interesting data and then plotting it.
211. Plug in the device.  Ensure that it is the only device plugged in
22   since this script is of very little brain and will get confused otherwise.
242. Enable the Window Orientation Listener debugging data log.
25   adb shell setprop debug.orientation.log true
26   adb shell stop
27   adb shell start
293. Run "orientationplot.py".
35The tool displays several time series graphs that plot the output of the
36WindowOrientationListener.  Here you can see the raw accelerometer data,
37filtered accelerometer data, measured tilt and orientation angle, confidence
38intervals for the proposed orientation and accelerometer latency.
40Things to look for:
421. Ensure the filtering is not too aggressive.  If the filter cut-off frequency is
43   less than about 1Hz, then the filtered accelorometer data becomes too smooth
44   and the latency for orientation detection goes up.  One way to observe this
45   is by holding the device vertically in one orientation then sharply turning
46   it 90 degrees to a different orientation.  Compared the rapid changes in the
47   raw accelerometer data with the smoothed out filtered data.  If the filtering
48   is too aggressive, the filter response may lag by hundreds of milliseconds.
502. Ensure that there is an appropriate gap between adjacent orientation angles
51   for hysteresis.  Try holding the device in one orientation and slowly turning
52   it 90 degrees.  Note that the confidence intervals will all drop to 0 at some
53   point in between the two orientations; that is the gap.  The gap should be
54   observed between all adjacent pairs of orientations when turning the device
55   in either direction.
57   Next try holding the device in one orientation and rapidly turning it end
58   over end to a midpoint about 45 degrees between two opposing orientations.
59   There should be no gap observed initially.  The algorithm should pick one
60   of the orientations and settle into it (since it is obviously quite
61   different from the original orientation of the device).  However, once it
62   settles, the confidence values should start trending to 0 again because
63   the measured orientation angle is now within the gap between the new
64   orientation and the adjacent orientation.
66   In other words, the hysteresis gap applies only when the measured orientation
67   angle (say, 45 degrees) is between the current orientation's ideal angle
68   (say, 0 degrees) and an adjacent orientation's ideal angle (say, 90 degrees).
703. Accelerometer jitter.  The accelerometer latency graph displays the interval
71   between sensor events as reported by the SensorEvent.timestamp field.  It
72   should be a fairly constant 60ms.  If the latency jumps around wildly or
73   greatly exceeds 60ms then there is a problem with the accelerometer or the
74   sensor manager.
764. The orientation angle is not measured when the tilt is too close to 90 or -90
77   degrees (refer to MAX_TILT constant).  Consequently, you should expect there
78   to be no data.  Likewise, all dependent calculations are suppressed in this case
79   so there will be no orientation proposal either.
815. Each orientation has its own bound on allowable tilt angles.  It's a good idea to
82   verify that these limits are being enforced by gradually varying the tilt of
83   the device until it is inside/outside the limit for each orientation.
856. Orientation changes should be significantly harder when the device is held
86   overhead.  People reading on tablets in bed often have their head turned
87   a little to the side, or they hold the device loosely so its orientation
88   can be a bit unusual.  The tilt is a good indicator of whether the device is
89   overhead.