Usage of a Pitch and a Roll angles without any road context would be redundant
Usage of a Pitch and a Roll angles without any road context would be redundant.
Splitting issue #380 (closed) - part 3
An Entity traveling along a road is deemed in the OpenSCENARIO as being one of the vehicular transport facility. That means the orientation of the Entity in the space is always determined by the underlying road surface. The Entity cannot have the orientation in terms of a pitch and a roll, which would be different and uncoupled from the road (in comparison to an airplane in the aviation domain). As stated in two places of the OpenSCENARIO User Guide (cf. Section 3.7 Trajectories):
Instances of Trajectory are used to define an intended path for Entity motion in precise mathematical terms. <...>
Alternatively, instances of Trajectory may also be specified using the three-positional dimensions and the three-rotational dimensions (heading, pitch and roll) for six total dimensions. In the second case, the path not only specifies the movement of the entity along the path, but also the orientation of the corresponding entity during that movement. <...>
So, a Trajectory is deemed as an intendent path (not a pose of an Entity in the space!). It looks curious with respect to the Pitch and Roll angles as an Entity (even with help of an assigned Controller) cannot position itself in the space freely, it is always limited by the road surface (only the Heading is under control). Here, we should exclude accidents from the consideration as the resulted position/orientation of an Entity would be out of control in such cases. We cannot prescribe an Entity to orient itself with a certain Pitch or Roll angle in the end. To be honest, a crashed vehicle turns into a MiscObject.
The same situation looks to me regarding a Pedestrian. The orientation (pitch and roll angles) of a Pedestrian seems to be redundant as well. Any pose of a Pedestrian could be defined by means of its Bounding Box dimensions. From the scenario perspective, it is obviously unimportant if a Pedestrian faces or turns side or back to the coming vehicle or even lies back. All poses can be modeled by the respective Bounding Box. And the response reaction of a vehicle should be the same.
On top of that, an ambiguous situation occurs when we try to apply a Pitch and Roll angles to MiscObject which is determined only by its Bounding Box model (there is no way to specify its "face" or "rear"). As its initial projection on a road is a rectangle, no problem would occur with the Heading angle, the resulted orientation would be the same. But when further rotations come into play, the result would dependent on which side is considered a "face". So, the full (h/p/r) orientation of a cuboid could be interpreted differently.
In practice (simulation), the Heading angle is widely used, what we cannot say about other angles.
As well, the OpenSCENARIO does not deal with a description of the space surrounding the Entity. Such data are mostly and in a very limited way borrowed from the OpenDRIVE file.
I'd like to raise the question from the title up to the discussion. Do the aforementioned arguments seem to be convinced to obsolete the Pitch and Roll angles in the Orientation class? If not, could anybody bring an exemplary scenario (XOSC file) from a practice where it was really used in the simulation with Pitch and Roll angles?
Please express your opinions.