Re: axes and normalization of orientation events

> With rotation, acceleration will be different at different points in the
> device.
True, but I still don't see the need to specify the origin of the
frame used for rotation. I think what's important is that the position
and velocity (from the Geolocation API) and the acceleration (from the
devicemotion event) refer to known points in the body, ideally the
same point. Assuming that the body is rigid, this is sufficient to
calculate the position, velocity and acceleration of other points in
the body under rotation.

We don't have any use-cases which which use rigid body kinematics to
calculate velocities or accelerations at arbitrary points in the body,
and hence require us to specify an origin, but I can see why this
would be useful.

So in Section 4.3 [1], we should probably update the wording 'must
provide the acceleration of the hosting device' to use a specific
point. The centre of mass seems like a good choice. Similarly, we
should update the Geolocation spec to make clear that the position and
velocity there refer to this same point.

[1] http://dev.w3.org/geo/api/spec-source-orientation.html#devicemotion

Thanks,
Steve

Received on Tuesday, 5 July 2011 12:40:09 UTC