next up previous
Next: Objective Models Up: Design Patterns for Avionics Previous: Decomposing ACS Models

Navigation Models


For purposes of avionics control, it's been found that the world of air navigation may be broken up into a few standard categories, with details that vary according to particular ``top-down'' properties needed to support guidance algorithms and ``bottom-up'' capabilities of sensor hardware. Navigation model properties may be needed at a number of points in an ACS, always including the computation of Error Models and, in many cases Guidance Mode Subsystems, where actual states are used as as initialization or correction data for desired state computation.

Design Steps

Organize navigation models around the four basic categories of the Aircraft, the Earth, the Atmosphere, and Transient Objects. Modeling the physical and control state of the Aircraft and its components is of course central to navigation. The geometric properties of the earth and fluid properties of the atmosphere must be represented because some navigation estimates are defined and/or computed relative to them. ``Transient'' objects include other aircraft, obstacles, and radio bases that are noticed and forgotten over time. All model components share a common structure, but necessarily differ in role and usage.

Aircraft. Some values represented by aircraft models are shown in the accompanying diagrams.

The collection of submodels representing an Aircraft forms a classic whole-part structure, paralleling the design of the aircraft itself. For the purposes of an ACS, an Aircraft contains composite attributes such as the overall position and heading, as well as a collection of component states associated with different parts of the aircraft (engines, rotors, wings, wheels, etc). But, as with most ACS models, there is no good reason for making a fixed, explicit model of the Aircraft as a single composite entity. Instead:

  1. Organize aircraft state into a group of independent submodels according to the workings and attributes of each part.

  2. Collect only those attributes common to the aircraft as a whole into components representing Central Aircraft State. This may itself be structured as a loosely coupled bundle of components, each maintaining a set of independent attributes or views.

  3. Aircraft subcomponent models most often play the role of estimation Stages, with data that are collected and combined in assessing central state. They should be structured and used accordingly. Associated Connections need be arranged only across subcomponent models needed to collect and reduce aggregate values across parts into single readings.

Earth and Atmosphere. While the earth and sky are pretty fascinating objects, to a navigation system, they are dreadfully boring. The main reason for modeling the earth is to make sure that the aircraft gets to where it is going, which is normally expressed as a relative position with respect to the earth. The atmosphere is modeled for similar reasons; for example to determine true speed given relative speed with respect to the air.

Earth and atmosphere models incorporate large amounts of static fixed knowledge about geometry and atmospherics. Models of the earth, especially, should be highly pre-wired with facts about the earth geometry, magnetic variation, etc. Updates mainly provide fine tuning. In most other ways, these models play the role of intermediate Stages, with data that are used as inputs along with others to ultimately estimate useful aircraft properties.

Transient Models. Only a few other real-world objects are of interest to a typical ACS. Some objects, such as Radio Bases, are statically known to exist, but are not always detectable because they enter and leave detection range. Updates and information from such objects usually form parts of earth models. Normally, all other objects (obstacles, threats, other aircraft) need be represented only with respect to fixed type categories and estimates of position and motion.

next up previous
Next: Objective Models Up: Design Patterns for Avionics Previous: Decomposing ACS Models

Doug Lea
Tue Mar 28 08:50:41 EST 1995