JeF,
There were quite a few screenshots in this thread and you did not specify which. I'll assume that you might have been refering to this one:
My reasons:
(1) As good as the AI is for path finding in RDOA/HTTR, I still prefer not to leave too much to chance.
(2) I'll specify very few waypoints when:
(a) The transit is likely to be fairly obvious and threats are low.
(b) I want the AI to find the fastest route and use it.
(c) Threats are high, but I personally don't have a good idea where they might be. So, I want the AI to respond to them as they are found.
(3) I'll specify some to many waypoints when:
(a) It's a long distance and I don't want the AI to get creative like taking a route that takes it through or near an objective. Even if Intel reports show no enemy contacts at or around an objective, prudence would dictate that an objective and nearby roads are a magnet for the bad guys. However, how is the AI to know? Because it might be a meeting engagement and thus, the enemy may not have had time to reach the objective. Or you might be conducting recon and your goal is to run across the enemy. Or you may suspect that the enemy is there but don't think they are present in sufficient strength to present a problem, but maybe time or positioning is of the essence.
In short, it's hard for the AI to know what is your intel picture (not the one it reports, but what you believe) of the battle and to know what your actual intent is when moving forces in the grand scheme of things.
(b) The engine does not permit changing route directives at different waypoints. So, I might want to do something like say while traveling from A to B avoid the enemy (since engaging them will only remove you from the battle) and while traveling from B to C force your way through (since you are needed where I told you to go and if you encounter resistance, then you still accomplishing something of value). Of course, you can perform multiple order cycles, but this also adds delays.
(c) I am intent on performing recon of a specific route and I don't want the commanded unit to find a better path. It's mission is to reveal the enemy on the ordered path even if that means that it will take horrendous casualties or find that it has no hope of reaching the specified end point.
(d) There are specific cases for which I tend to specify an extreme number of waypoints as in the screenshot above. Two which come to mind are:
Case I: I wanted the attackers to make their way to the area of St. Oedenrode as fast as possible. However, regardless of any intel reports, I strongly suspect a German presence at Ruzingen. Now, I want to FUP North of St. Oedenrode and hopefully unobserved by the Germans. So, I want to get off the road before I hit Runzingen. Using fewer waypoints, I run the risk of the AI plotting a fast route to my FUP that might take me across enemy positions (as I did want speed for 95% of the route, but not here). By using so many waypoints, I am overriding the route finding AI and leaving it no choice but to simply take a direct path between each and therefore implement my desired route.
Case II: Often I may want to set up particular units along a tree line or along the edge of a BUA to cover a road/clearing. I find that generally the AI will loop around the front of the woods or BUA to setup to cover the road/clearing. This fine and good when you are setting up a prepared defense. However, in a hasty defense when the lead is already flying this is somewhat suicidal and results in unnecessary casualities. So, I will specify a detailed route to the position which keeps cover between the unit and likely avenue of enemy approach. Granted I could try setting the route type to SAFEST or COVERED (which I have rarely experimented with), but once again my orders were to the ATG unit something like "get to the woods as fast as possible and then deploy to the tree line while avoiding unnecessary casualties".
---
JeF, I hope that I have answered your question satisfactorily. On the whole, I find the AI path finding in HTTR very good. It saves a lot of time and adds to the high-level feel of the game. For those who have not played RDOA yet, setting waypoints is nothing like what is required in flight sims and tactical combat games like Combat Mission. Often just a faint suggestion of the route you desire is sufficient to have your AI underlings work out the details. Also, you will commonly depend on your AI subordinates to respond to unexpected situations as they develop both in terms of path finding and providing security for the task force while it is in transit.
So, bottom line the AI path finding in HTTR is impressive, but it still doesn't read my mind. [:)]