What are the basic assumptions and simplifications? [14-05-03]

In the following paragraph assumptions and simplifications about the way the bot issues have been addressed are detailed. For sure, some information will be missing, please let us know.

The first part deals with simplifications made on the scenario itself, that is to say, the ‘characters’ and their roles. The knowledge section is about what assumptions were made to address the missing information issue. The last part talks about the simplifications for the decision making process.

Scenario

The program that will be implemented only simulates the behavior of the attacker aircraft in the dogfight. The geometry constraints prevent us from dealing with any reversal situations. As soon as the attacker is almost overtaking the defender, that is to say, as the attacker is almost becoming the defender itself, geometry constraints are violated so that the program does not provide coherent outputs anymore.

Knowledge

As some know-how skills are missing, many abilities such as computing the acceleration to reach one point from another one have been implemented by setting data at random. For instance, the AI layer provides what it wishes via goals it wants to reach. It is up to the layer below (know-how layer) to find out how to reach these goals. The AI layer provides the wanted speed speed (t1). The know-how layer knows the initial speed speed (t0) and speed (t1). The acceleration can be computed following this approximation.

This issue of missing skills is discussed further in the answer to the question: How has the missing information problem been fixed temporarily?

Decision Process

• Input simplifications:
• To compute decisions only the positional aspects of the inputs have been taken into consideration. In the prototype no energy considerations have been taken into account. For instance, only range and angles are processed to provide ‘maneuvers’. A low speed that means that the aircraft kinetic energy is dangerously low does not influence the decision making process.

• As soon as you launch the program, the defender aircraft is performing a turn. It is the only kind of enemy’s trajectory that is currently valid for the decision process as input.

• Output simplifications:
• The output maneuvers deal with more or less tight turns only.

• Process simplification
• In the prototype, the decision making process does not take into account any predictions on the enemy’s trajectory. For example, let us suppose that at t0 and t1, the AI module provides a maneuver. It is assumed that between t0 and t1 the enemy is still turning the same way. At t0 there is no predictions about what will happen at t1 except the fact that the enemy is turning with the same turnRate.
• It is assumed that all the information is obtained. None is missing. Right now, all information is required so that the conditions of the decision tree can be checked and the rules fired. Other solutions could be dealt with but time will miss.
• Decision Making Domain of validity:
• There are some geometry constraints. Beyond these limits, the behavior of the attacker is not necessarily coherent. See the answer to the question for more details: Which situations should the defender aircraft face?
• Aircraft are flying in the plane. Right now, neither inputs nor outputs take altitude into consideration.