An idea for how to make AI co-pilot track a piece of ground/feature, which is not an object

[zerO_crash]

Regarding the types of missions and tasking that can occur in a given mission, it is not always possible to direct the AI co-pilot on a given target (ground feature, default building on the map, etc...). I have been thinking about how this could be solved in an elegant way, and came up with the following:

 

The pre-requisite for this to work properly, is that only manual fire will be accepted (player pulls the trigger). Otherwise, if AI is to commit to engaging, the pilot would have to have a command/action for forcing through the AI's decision on whether to engage or not (hold the trigger for longer, or something similar - that while in auto-mode (not manual)).

 

As to the mechanic, when the pilot wants the AI co-pilot to scan a certain area, he/she places the imaginary reticle at the spot, and gives the order to the co-pilot to inspect that area. What follows after, if objects are spotted, is the pop-up of a contextual menu with target selection list. Here is where my idea for a change comes into play. Why don't we improve that menu, by creating one "default" point at where the cross was when the order to inspect was given? (This is in essence, a SPI-point). Whenever the order to inspect is given by the pilot, the first option to always show on the list, will be "point". If objects are spotted additionally, the list would be populated, just like it works currently. If no objects are spotted, the "point"-option will appear alone. In any case, it would grant the pilot full antonomy with which to chose targets, and not only objects, but just about anything physical (ground, civilian vehicle, default building on a map, etc...). It's really just a matter of just adding the initial point of interest (where the reticle is when the order to inspect is given) as a "targetable" point.
 

This idea, would build further on the mechanic, and be easy and quick to use in nature, especially in dynamic scenarios. It could be applied to other modules as well.

 

zerO

Edited January 28, 2024 by zerO_crash

[admiki]

I like your idea, but aiming with your head at something that is few clicks away is inprecise to say the least.

What I would like to have from ED is an option for mission designer to add a point on the map that will be objectified into a target as current units are.

[zerO_crash]

I was considering that, and whilst with a bit of training, one should be able to place the marker with a relative accuracy, at the end, the further the distance, the greater the inaccuracy. As to the map, it could work as well, but then you miss the dynamics of using this in combat. Also, the precision does get lost, as from the map, you'd be guesstimating on the position of the target. Consider a non-object somewhere in the middle of a farm field.

 

There is bound to be inaccuracy with the system nevertheless. In order to mitigate it with my method, altitude and distance are the main factors to consider, when designating, especially if precision is of essence. The one additional mechanic that could be added here, would be a hat-command (preferably layered from something already being used) to give the co-pilot commands in the sense of "Further up", "A notch to the right", etc... It could definitely allow for increased precision at range, if one prefers this over creating a new point. In either case, having the "point" selected, you would judge based on where the Raduga-Sh is looking (cross on pilot's sight).

 

To be quite honest, this would permit us to use bombing, and other less accessible features or with decreased accuracy (without a human co-pilot), finally with AI, and efficiently at that.
 

Off the record: The solution with a hat, allowing for incremental adjustments, would seemingly work well with something like e.g. an Apache or F-14 too, as there you can observe what co-pilot/WSO is looking at on a screen. If you need to move that point, give a couple of commands with a hat-switch (talk on), and off you go. The reason I mention other modules, is that the system should be uniform across modules. That way, it remains a learn-it-once and becomes more accessible to most.

Edited January 28, 2024 by zerO_crash