Droids movement more accurate

Ideas and suggestions for how to improve the Warzone 2100 base game only. Ideas for mods go in Mapping/Modding instead. Read sticky posts first!
User avatar
Rider
Trained
Trained
Posts: 106
Joined: 31 May 2007, 16:06

Droids movement more accurate

Post by Rider »

This is something that's bothered me since the original game came out and I really don't know why I never brought it up before :)

Droids are not accurate in achieving their move orders. Ordering a droid to a position on the map will make it get pretty close, but if you intend on setting them on pretty specific spots, or in a particular orientation it becomes an exercise in futility. For defensive units to face a certain way can actually give you an advantage, particularly with slow turning turrets. Then there's the "Nudging", where friendly units move in groups, are ordered to stop and then continue bumping into eachother for a bit trying to find a position to settle down. This is particularly annoying when units move near pre-placed units and they get bumped into, and thus move away from any advantageous orientation you may have set them up in. Also it affects the auto-repair as that only works when a unit is still.

A large contributor to this problem, I think, is the droids limited turning circle. They're unable to turn in place or make any kind of sharp turn because of it. This is particularly odd for tracked units as tracks are supposed to allow a unit to rotate around it's own axes.

I hope this is something that'll get some attention in the (near?) future, I think it can make the game feel more 'tight' because of this extra degree of control, and I personally would love to see that in WZ2100 :)
User avatar
Zarel
Elite
Elite
Posts: 5770
Joined: 03 Jan 2008, 23:35
Location: Minnesota, USA

Re: Droids movement more accurate

Post by Zarel »

...that's not a feature request, that's a bug.

I just recently started taking a look at the droid movement code. I'm currently working on fixing it, but I can't seem to figure out how.
kringled
Trained
Trained
Posts: 137
Joined: 16 Jan 2010, 16:53

Re: Droids movement more accurate

Post by kringled »

A particularly dramatic example of this is when I have a large group (under a commander) of all hover units, which are particularly prone to nudging. I've seen them slowly drift across large portions of the map, including up slopes and ending up in a contained valley.

Keith
Seismo
Trained
Trained
Posts: 70
Joined: 06 Feb 2010, 17:32

Re: Droids movement more accurate

Post by Seismo »

Zarel wrote:...that's not a feature request, that's a bug.

I just recently started taking a look at the droid movement code. I'm currently working on fixing it, but I can't seem to figure out how.
Well accuracy belongs first to the tiling size of the map. The second thing is that positioning should be done with float instead of int values ;)
User avatar
Rider
Trained
Trained
Posts: 106
Joined: 31 May 2007, 16:06

Re: Droids movement more accurate

Post by Rider »

Maybe make use of something like SupCom2's Flowfield Pathfinding?
Seismo
Trained
Trained
Posts: 70
Joined: 06 Feb 2010, 17:32

Re: Droids movement more accurate

Post by Seismo »

wow, thats how the things should run, very nice.
User avatar
Zarel
Elite
Elite
Posts: 5770
Joined: 03 Jan 2008, 23:35
Location: Minnesota, USA

Re: Droids movement more accurate

Post by Zarel »

Hah, like we have developers who would implement something like that for us. :P
User avatar
Rider
Trained
Trained
Posts: 106
Joined: 31 May 2007, 16:06

Re: Droids movement more accurate

Post by Rider »

Well not 'for' you per se... however, given that this issue would have a high enough priority, I'm sure someone already on-board could be persuaded to look into it?

A lot of the research into this type of path finding has already been done apparently. There's, for instance, this document on direction maps, which would support a flow field approach to pathing. There's also this (fugly) site which touches on the subject of flow fields and general path finding. In essence the research is already there, you'd just need someone to make sense of it and transform it into usable code.

It'll take time, sure, it just depends if this will justify that time in the end. I personally feel that it does, but I'm not the one implementing it, so take that for what it's worth :)