Factorio Forums

I think rather a lot of players would like to see certain aspects of the game improved, such as more intelligent logistic bots. These requests are typically rejected as they would decrease UPS performance. This got me thinking- what if the game had two sets of behaviours, one for smaller bases and one optimised for performance? If the game started to drop below 50 ups it could swap to the optimised set.

Just a thought, try not to crucify me too vigorously

Every example of needing smarter bots i've seen is a result of poor design. Design your base so that you are using the bot behaviours, rather than needing different behaviours.

Yes. Don't make U-sized logistics networks. Don't have production 20 miles away from consumption. Use buffer chests. Have enough roboports. If you design it right, better pathfinding wouldn't be needed.

The bots idea... I don't agree with, sorry
The idea of adjusting performance, however, would be quite good late game, due to the high power involved.

Jap2.0 wrote:Yes. Don't make U-sized logistics networks. Don't have production 20 miles away from consumption. Use buffer chests. Have enough roboports. If you design it right, better pathfinding wouldn't be needed.

What is a U-sized logistics networks ?

I'd like to see someday RTS style Modification for factorio. Maybe even some kind of RTS official expansion. But in this RTS style-thing wee need more features to manage large bot networks. We don't need heavy, comlicated AI-thing for bots. Just little things: filters for storage chests, ability to bind bots to specific roboport, or roboports (i think this feature can be UPS friendly? Maybe if binded bots just ignore the existence of other roboports), maybe also filters for bots, or next tier: filter bot, so you can have bot fleet only for iron plates, another fleet only for copper plates etc.) Of course bots nerfs and belts buffs are necessary to make this work, because in current state such features may lead to the definitive death of belts after bot research.

But whatever, devs have ATM too much work to do to even think about such things. So i just shut the fu** up. So much douchebags think that their idea is the best and most important in the world, so dev's should implement that ASAP, i don't wanna be like them

userasd wrote:

Jap2.0 wrote:Yes. Don't make U-sized logistics networks. Don't have production 20 miles away from consumption. Use buffer chests. Have enough roboports. If you design it right, better pathfinding wouldn't be needed.

What is a U-sized logistics networks ?

When you have roboports in a shape so that when robots try to go from one side to the other, they run out of power and go back to where they came from to charge, and get stuck in a loop.

userasd wrote:
What is a U-sized logistics networks ?

That is a logistic network in shape of an "U".
Maybe you have you're base on bottom and on left side a flank to iron deposit and on right side a flank up to copper deposit.
In between are some alien bases.

Now a bot try to go from copper-side to iron-side will not be routed over base, it will try to go straight trough alien territory.

I'm no native English speaker. But I think but I think it should named as "U-shaped logistics network" ...

TheRaph wrote:

userasd wrote:
What is a U-sized logistics networks ?

I'm no native English speaker. But I think but I think it should named as "U-shaped logistics network" ...

Yes, I worded that poorly.

Thanks, guys! :3

I highly doubt developers will implement something like this. Why?
Adding another different game behavior would double the code needed for game as each behavior set would have its own code. Also doubling the code would also mean doubling the chance for new bugs being introduced into the game and not to mention much harder way to track specific bug.
So all in all it would require lots of work for little benefits. Therefore it is better that developers spend all that time to optimize existing game behavior. But due to the very high complexity of the game and highly complex bases that players manage to build this is far from begin an easy task.

How much complexity would it add, for example, to tweak personal roboports to build from closest object first, unless a performance limit is hit? It'd seem like a relatively simple change and easy to add a little tag somewhere saying which behaviour pattern is in use. That way you could still have simple QOL improvements in singleplayer which don't impact the game's ability to handle a big MP session.

In retrospect logistic robots were a bad example because a base which worked previously might not once the lag killer kicks in, depending on the changes made (I was thinking more of closest robots are assigned a task instead of random). Regardless, I appreciate that I don't really know what I'm talking about from a coding perspective and just wanted the idea to be out there.