W3C home > Mailing lists > Public > public-geolocation@w3.org > May 2011

Re: axes and normalization of orientation events

From: George Percivall <gpercivall@opengeospatial.org>
Date: Mon, 16 May 2011 17:47:57 -0400
Cc: Steve Block <steveblock@google.com>, public-geolocation@w3.org
Message-Id: <C469B6CC-1DFD-4B3B-BF73-D572EEB746C7@opengeospatial.org>
To: Dean Jackson <dino@apple.com>

Did not intend to confuse.   Seems two use cases should be discussed: e.g., map viewing, object dynamics. 

My comment about looking down on a map was based on Steve's comment:
>>> Talking about clockwise vs counter-clockwise only makes sense when you specify the direction the observer is facing.
Which is absolutely true. 

As Steve, said the diagram in the spec uses "East - North - Up" (ENU):
ENU is a good choice because it is a right hand rule coordinate system.
ENU is a bad choice if you want to be consistent with typical practice for maps and vehicle dynamics.

For automobile dynamics consider SAE J670: Vehicle Coordinate System.
SAE charges for their standards, but you can see the relevant figure here:
And SAE J670 is normative reference in J1733 (see figure 3), which you can find on the web here:
SAE J670 defines  X forward in the vehicle, Y out the side of vehicle, and Z down which can be called "North East Down" (NED).

The OGC Snapshot spec uses NED as the spec is designed for both maps and vehicle dynamics.
- For maps its consistent with heading when looking down at a map; heading clockwise from north.
        This seems relevant as the figure in DeviceOrientation implies orientation on a display, which might be a map.
- For vehicles, its consistent with J670 and other standards.

My comments are relevant if these use cases (maps and vehicles) apply to the Device Orientation spec.


On May 16, 2011, at 3:04 PM, Dean Jackson wrote:

> OK, now I'm even more confused :) These diagrams are not a map, but should we follow that convention?
> The good news is that I think our (Apple's) existing implementation is correct, and Steve's explanation of the current diagram makes sense. I'm also fine about the choice of X/Y axis direction - I only pointed it out because 90+% of the documentation on the Web *is* related to vehicle dynamics. I didn't find any documentation with orientation defined for a software application, so this might be newish ground.
> Dean
> On 17/05/2011, at 1:42 AM, George Percivall wrote:
>> Steve, Dean,
>> Typical practice is to look down at a map.
>> Typical practice is to measure heading clockwise from north while looking down at the map.
>> So, the coordinate system using right hand rule for this approach is North - East - Down. 
>> The OGC Moving Object Snapshot uses NED in this way:
>> - Local Tangent Plane (Section 5.3)
>> "The velocity in MovingObjectSnapshot is defined relative to Euclidean plane corresponding to a Local Tangent Plane (LTP) at the location of the object.  The LTP is defined with its origin at the point where the object is located on (or possibly near) the Earth’s surface. LTP is oriented in three dimensions with the vertical axis taken to be straight down, parallel to the gravitational gradient, with the plumb line. The other two axes are perpendicular to the vertical axis aligned with local geographic north and east."
>> - Heading" (Section 7)
>> "...value must be in degrees measured clockwise from due North"
>> Regards,
>> George
>> On May 16, 2011, at 11:16 AM, Steve Block wrote:
>>> Hi Dean,
>>>> However, in researching it I notice that the W3C spec lists the axes in a different order than the convention used by
>>>> the rest of the world.
>>> I'm aware that in vehicle dynamics the x axis usually points in the
>>> direction of motion, but do you have a reference for a standard for
>>> non-vehicle applications, like a UA? We chose the Earth frame to be
>>> East-North-Up and it made sense to align the body frame such that it
>>> matches the Earth frame when viewing a map on the device, though I
>>> don't feel strongly about this.
>>>> Also, I notice that the spec says "Rotations use the right-hand convention: positive rotation around an axis is
>>>> clockwise when viewed along the positive direction of the axis". It's highly likely that I'm misinterpreting this, but a
>>>> rotation around Z using the right hand rule (from X axis to Y axis) is counter-clockwise. In fact, the little arrow on the
>>>> diagram shows this.
>>> Yes, I think you're misinterpreting this. Talking about clockwise vs
>>> counter-clockwise only makes sense when you specify the direction the
>>> observer is facing. +ve rotation about an axis appears clockwise when
>>> the observer is facing the direction of the +ve axis. Conversely, if
>>> the observer is facing the -ve axis direction (as is the case for the
>>> z axis in the second diagram), the rotation appears counter-clockwise.
>> George Percivall
>> gpercivall@opengeospatial.org
>> @Percivall on Twitter
>> +1-301-560-6439
>> http://www.opengeospatial.org/
>> OGC -- Making Location Count

George Percivall
@Percivall on Twitter
OGC -- Making Location Count

Received on Monday, 16 May 2011 21:48:22 UTC

This archive was generated by hypermail 2.4.0 : Friday, 17 January 2020 19:51:02 UTC