Friday Facts #224 - Bots versus belts

[keyhenge]

Having gone through about 22 pages, I haven't seen something resembling my opinion yet, so I made an account.

I don't think the problem with belts and bots is one being overpowered; it's a matter of complexity and progression. As it stands, bots can do everything belts can do, and do it better. This will just about always be the case because of how bots and chests currently work (eliminates the need for land management and sorting), unless there are massive nerfs to both of these items which I don't think anyone will like. Instead, I think bots should be thought of as the "next step" past belts; the automation of space management, and sorting, while bringing their own new challenges. Then, the problem isn't that bots are overpowered, it's that the logistics system the bots use doesn't bring enough challenges to be interesting.

Consider the following situations of adding to a complex network of belts, trains, and then bots. Assume a tightly packed factory for each.
In order to add to the belt network, parts of the factory need to be re-planned from scratch. This is becomes more interesting as more efficiency is needed, as you need to consider a wider factory around you constantly limiting your space.
In order to add to the train network, the timing of your currently existing trains need to be considered if you're staying with your currently operating stations. If you're making a new station, the factory as a whole limits your space (or the train might limit the factory, either or). As you add more and more trains, space runs thin and timing becomes harder.
In order to add to the bot network, you plop a new assembler with an input and an output. This unfortunately barely interferes with the rest of your factory since new expansion can always be handled this way. If there's a problem with throughput, you make more bots and roboports.

The logistics system doesn't add any new complexity as factories become larger in the same way train networks and belt networks do. It's mostly because of how incredibly streamlined the logistics system has become. Barring the initial confusion of how to work it, logistics is fairly simple; provider chests for any output, requester chests for any input, and storage chests as a buffer. All of these operate on the same network, and individual chests aren't really treated differently. To expand on that, storage chests really just form one big supply from which bots can take; same for passive provider chests. Active provider chests send to this supply or to any open requester chest. There aren't player decisions being made here, it's essentially just become a game of increasing and balancing throughput through hard numbers. In my opinion, the logistics system should resemble something like programming; items are values of certain types stored in variables (chests) which need to be delivered in a consistent and fast manner. I'm not sure as to how it would work in its entirety (others can add their ideas), but I think that making the logistics system more complicated falls in line with what the game is about in the first place, that being increasingly complex logistic problems solved by further and further automation.

[rldml]

Trains are actually a great example of balance because despite having massively increased throughput compared to belts, they have their own pretty hefty limitations that keep belts relevant. The proof is there - high end bases are very often 100% bots with few or no belts, but 100% train bases with no belts are a novelty at best.

ATM:

Code: Select all

TYPE        USAGE                                                THROUGHPUT
Trains      Bulky, clunky mechanics                              High
Belts       Flexible but requires a fair bit of space/planning   Medium
Bots        Little/no real estate, ultimate flexibility          Medium-High??*

*Shouldn't that be low? Don't remove them or nerf to the ground, I just think there's a fair point that bots are too dominant.

That's a good base to think a step further: Make bots transport only one item at the time and make them more power consuming. I believe, you need them to nerved to make belts not obsolete in endgame.

Advantages:
- To transport the same amount of items with bots you would need twice of them as now.
- Greater power consuming reduces the amount of robots you can support at one time.

In general, you would decrease the throughput in a significant way.

[malecord]

My two cents. Logistic bots are evil only because they are not "realistic". In the sense that they occupy no space and so are collision free. And this allows them to defeat physic and achieve impossible throughput. You can buff belts as much as you like but they will never match logistic bots throughput simply because bots have no throughput limits. They scale infinitely in addition to be extremely flexible (as they should be).

Give them collision box, force them to respect safe distance from each other and you will have brought them in the factorio spirit where you have to take in account numbers and designs. They will still have their use, be part of smart Logistic designs but they won't be anymore the brainless solution for all the problems in the game.

"Give them collision box" and thus make them hell on UPS and game logic. Your performance, even with just a few dozen would suffer. A couple hundred? Bye bye, UPS.

This would be the worst possible nerf because they'd be unusable (not to mention it wouldn't be an easy fix to do, it would surely be a lot of work).

You fail to see the issue.

Bots ignore space. One of the fundamental elements of Factorio is 2 dimensional puzzling. You have to figure out how to organize your space and to accommodate belts, inserters and machines so that the stuff is done. Bots don't change this paradigm, they broke it completely because they occupy no space at all. It's not like "I have to solve this puzzle differently", it's "I have no puzzle to solve". Everything in factorio has some space requirement and/or area of effect constraint that combines into a into a 2D puzzle to solve even when it doesn't make any sense like for beacons (if they went the sensible way they would have made them so that they require a cable connection with the assembler).

Collision boxes is how real drones works. But this is just a game. You just need an approximation that gives you a space constraint to force you to solve a 2D puzzle (in real world it's a 3d puzzle anyway). Somebody suggested to add a loading animation and make so only n bots can load/unload stuff from a chest at the same time. This is an example of 2D puzzle. To scale you need to add more chests and distribute stuff efficiently. Another possibility is to limit the max number of concurrent requests handled per roboport in their area of effect. Like only 5-10-15 per roboport level / tech upgrade at the same time. This would require you to add more roboports and/or spread your logistic chests on more space to avoid saturation. Another 2d puzzle example.

Both example these don't require individual bots collision boxes but they approximate them by limiting how many bots can operate on or near to the ground at the same time. That's more than enough to make them fit into the puzzle theme without breaking them completely. Plus they could be easily modded to behave without constraints like it is now.

[Drury]

I don't feel strongly about the balance of the bots. I outright have no idea how good they are firsthand. I never use them. You can buff them or nerf them, I don't care. Here's why:

Seeing all the bots zip around... It's kinda cool, but it feels more like you're watching Wube's handiwork than your own. As far as you're concerned, all you did was place down some colored chests, some roboports, and filled them up with robots. There's no splitters or signals or complex circuitry, there are no interesting designs to come up with. Belts, trains, heck, even pipes are incredibly more in-depth than bots. The way you lay them down is unique to you - even if you copypaste those designs with blueprints and have them reconstructed by bots over and over, they are still your own. They give you a sense of pride and accomplishment, so to speak. Bots just solve all your problems in the blandest way possible.

So that's my main issue with bots. I don't care how balanced or practical they are. I don't care how much power they consume, or how much cargo they can carry. They could carry mountains, or barely move a speck of dust, I'll never use them for anything because they're boring. I feel the same way about the artillery to be honest, but that's for another discussion. (at least artillery is kinda good from a pure badassery standpoint)

If we want to fix the bots, we have to look at how they compare with belts and such in terms of creative freedom they allow the player and try to increase that rather than meddle with arbitrary values. Perhaps make bot networks more interesting than just overlapping logistics areas, make air traffic flow more of a thing to control somehow. If not that, add an entirely new sidegrade system to bots with the throughput and versatility of bots, but still with a design challenge to offer such as the pneumatic tube suggestion. People are suggesting buffing the belts - I'm not sure that'd help. Belts kinda become unwieldy past the midgame. You have so many different types of items that you need a separate belt for and the design challenge becomes a bit too much. Bots are attractive for this precise reason - you can just dump all of your many different items into the logistics network and it sorts them out automatically. With a belt, you need a spaghetti strand for each type of item and it gets tedious. A new transport system that combines the design challenge of belts with the mass item sorting capabilities of bots would be great.

EDIT: Essentially, do with lategame transport what you did with lategame power. Spamming solar+accs is the equivalent of spamming bots, adding a more complex nuclear option would be equivalent to adding the pneumatic tube. Depending on the players' preferences, they can either take the simple option or the complex option.

[sicklag]

.

[Chronosfear]

Bots ignore space. One of the fundamental elements of Factorio is 2 dimensional puzzling. You have to figure out how to organize your space and to accommodate belts, inserters and machines so that the stuff is done. Bots don't change this paradigm, they broke it completely because they occupy no space at all. It's not like "I have to solve this puzzle differently", it's "I have no puzzle to solve". Everything in factorio has some space requirement and/or area of effect constraint that combines into a into a 2D puzzle to solve even when it doesn't make any sense like for beacons (if they went the sensible way they would have made them so that they require a cable connection with the assembler).

Collision boxes is how real drones works. But this is just a game. You just need an approximation that gives you a space constraint to force you to solve a 2D puzzle (in real world it's a 3d puzzle anyway). Somebody suggested to add a loading animation and make so only n bots can load/unload stuff from a chest at the same time. This is an example of 2D puzzle. To scale you need to add more chests and distribute stuff efficiently. Another possibility is to limit the max number of concurrent requests handled per roboport in their area of effect. Like only 5-10-15 per roboport level / tech upgrade at the same time. This would require you to add more roboports and/or spread your logistic chests on more space to avoid saturation. Another 2d puzzle example.

Both example these don't require individual bots collision boxes but they approximate them by limiting how many bots can operate on or near to the ground at the same time. That's more than enough to make them fit into the puzzle theme without breaking them completely. Plus they could be easily modded to behave without constraints like it is now.

I like the idea of limiting the amount of robots per chest, and that they need some time to grab the goods and unload it elsewhere ( you could also add some level 1-3 or something increase loading/unloading speed and/or number of bots per chest)

Also I was thinking about adding a belt speed research for late game ( will improve every belt type by a % ) so maybe red belts would be then fast enough for smaller production lines ( would increase the ratio of cost to benefit belts if that may be something )

I don't use bots much because they make the game to easy and i love the view of items traveling on belts. I only use them for recipes that require stuff all around my base ( If missing room to lay a nice looking belt ) and of course constructionbots because you cant play endgame without them.

greets
Chronosfear

[ratchetfreak]

There is a big disadvantage to bots: debuggability for optimizations (I mean in a base-planning way)

when a bot base is badly setup it is very hard to figure out that it even is bad.

A bad base will have bots being empty for large parts of their travel time and/or making large detours for recharging or getting stored.

Figuring out if your base has these issues is a lot harder than finding out whether a belt is uncompressed or backed up.

[Ghoulish]

I don't think nerfing bots is the way forward, but rather buffing the speed of belts (and inserters to match). So research to increase the speed of an express belt, along with inserter speed buffs. Twinsen also mentioned recently about belt stacking which I find an intriguing idea, and I would guess very UPS friendly as most of the items in a stacked belt would be obscured? Another item that I think should be added for end game belts is the loader, it's hidden in the game as default, adding it along with belt and inserter speed buffs would be a nice step forward!

[agmike]

You fail to see the issue.

Bots ignore space. One of the fundamental elements of Factorio is 2 dimensional puzzling. You have to figure out how to organize your space and to accommodate belts, inserters and machines so that the stuff is done. Bots don't change this paradigm, they broke it completely because they occupy no space at all. It's not like "I have to solve this puzzle differently", it's "I have no puzzle to solve". Everything in factorio has some space requirement and/or area of effect constraint that combines into a into a 2D puzzle to solve even when it doesn't make any sense like for beacons (if they went the sensible way they would have made them so that they require a cable connection with the assembler).

Collision boxes is how real drones works. But this is just a game. You just need an approximation that gives you a space constraint to force you to solve a 2D puzzle (in real world it's a 3d puzzle anyway). Somebody suggested to add a loading animation and make so only n bots can load/unload stuff from a chest at the same time. This is an example of 2D puzzle. To scale you need to add more chests and distribute stuff efficiently. Another possibility is to limit the max number of concurrent requests handled per roboport in their area of effect. Like only 5-10-15 per roboport level / tech upgrade at the same time. This would require you to add more roboports and/or spread your logistic chests on more space to avoid saturation. Another 2d puzzle example.

Both example these don't require individual bots collision boxes but they approximate them by limiting how many bots can operate on or near to the ground at the same time. That's more than enough to make them fit into the puzzle theme without breaking them completely. Plus they could be easily modded to behave without constraints like it is now.

Then the whole game is overpowered and creeped. Let me explain.

Remember the gameplay when you just generated the map, the early game. You have a couple of machines, you personally interact with every and each of those. You placed manually every one of those, and each placement was a singular case decision.

Then you progress into mid game. You do not think in singular machines anymore, you think in builds as a set of machines consuming something and outputting a belt of something. You don't think of the placement in terms of machines, you place whole builds. You don't interact with machines, you get items from belt or buffer chests. You have therefore abstracted from the singular machine perspective, you think in terms of larger scale constructions.

Then you progress into late game. You think of your base as a whole. Maybe you even have multiple bases (main plus outposts?). You again have abstracted to a different perspective.

Then comes the end-game. You have researched all non-infinite research, scaling up the production. You place bases dedicated to concrete responsibilities, or self-contained builds, big constructs. You don't think about about fitting that output inserter for that assembler in that row inside that build. It is already solved, done, prepared, ready to be placed multiple times. You do not care about this puzzle. The puzzle is now on your M view — fitting that outpost, laying rail, ensuring total throughput, large scale logistics. You don't have time to think about that inserter anymore. It doesn't even matter if you use belts or bots in these builds — it would look the same, placing down the blueprint. There is no such puzzle anymore. The build is basically a black box. You have abstracted again.

This is how everything works, in real life too. When you zoom out, you lose sight of small details, because we are only human, and our minds can only handle this much. This is completely natural thing, what in Factorio we call progression. When the old puzzles become irrelevant, solved, they disappear, but new ones come in their place.

Now, you were talking about beacons and space efficiency and how bots synergize with it. However, you are wrong in that the underlying issue here is not the simplicity of the bots which makes megabase people prefer bots. It is the ultimate limitation of Factorio, which is not space, not complexity, not resources, it is your computer's ability to run the game. The reason why beacons and modules are heavily used is because they are performace efficient compared to just building more. The bots are used because they were more efficient than belts. The simplicity argument is just not there. If belts would allow 15k SPM, there absolutely would be belt based megabases doing 15k SPM, and complexity is not a stopper here. In fact, I would dare say that in vanilla there is no such recipe which is impossible to handle with belts in beacon-assembler-beacon sandwich style. Some maybe trickier than others to design, but this design is only a puzzle once. After that there is no difference between belts and bots in building it.

[vampiricdust]

I don't think the problem with belts and bots is one being overpowered; it's a matter of complexity and progression. As it stands, bots can do everything belts can do, and do it better. This will just about always be the case because of how bots and chests currently work (eliminates the need for land management and sorting), unless there are massive nerfs to both of these items which I don't think anyone will like. Instead, I think bots should be thought of as the "next step" past belts; the automation of space management, and sorting, while bringing their own new challenges. Then, the problem isn't that bots are overpowered, it's that the logistics system the bots use doesn't bring enough challenges to be interesting.

No, bots cannot do everything belts can do better. Belts are cheaper, have more consistent throughput, function whether you have enough power, can be built across an infinite number of tiles without (ignoring computer limits) losing throughput, and they do require you to build more power generation that either consumes more fuel or massive solar fields (96 panels needed to charge for every 4 bots charging). Bots are far more expensive, require an infrastructure, special chests, and logistics controls to keep them from being even less efficient. Bots are easy, but they are not good in the vast majority of uses.

Consider the following situations of adding to a complex network of belts, trains, and then bots. Assume a tightly packed factory for each.
In order to add to the belt network, parts of the factory need to be re-planned from scratch. This is becomes more interesting as more efficiency is needed, as you need to consider a wider factory around you constantly limiting your space.
In order to add to the train network, the timing of your currently existing trains need to be considered if you're staying with your currently operating stations. If you're making a new station, the factory as a whole limits your space (or the train might limit the factory, either or). As you add more and more trains, space runs thin and timing becomes harder.
In order to add to the bot network, you plop a new assembler with an input and an output. This unfortunately barely interferes with the rest of your factory since new expansion can always be handled this way. If there's a problem with throughput, you make more bots and roboports.

Space is nearly infinite as far as we are concerned. Making a tight base is a choice you make. Space things out more and belts are insanely better (except for UPS limitations and beacon spammed production). If you're having throughput problems, you should use belts, not more bots. If you think spamming bots is the answer, you're completely ignoring the issues causing the throughput problems. Bots have to travel from their starting roboport to the source chest, then to the end chest, diverting for power whenever needed, and then heading back to the nearest uncrowded roboport to top off and "land". A belt will always exist at the start & end points and move items instantly from point A to point B regardless of power, distance, or logistic coverage. For every roboport you make, that is ~13.5 blue belts of resources, for every logistic bot it is ~3.5 blue belts. It takes 8 logistics bots at 5 capacity per second to move as many items as every single blue belt does by itself. At 100% speed, logistic bots would move 3 tiles in 1 second, for 8 bots you could have made ~27.8 blue belts which move items 5.625 tiles per second. So at 100% speed, a blue belt would have moved 40 items 5.625 tiles while 8 bots would have gone only 3. It isn't until you research worker speed 5 that logistics bots start being faster. It would take Worker Speed research level 20 for 1 bot to move 8 times the speed of a blue belt to be "better", but even then it has to be about 3.5 times better to have better resource efficiency just straight up cost....

The logistics system doesn't add any new complexity as factories become larger in the same way train networks and belt networks do. It's mostly because of how incredibly streamlined the logistics system has become. Barring the initial confusion of how to work it, logistics is fairly simple; provider chests for any output, requester chests for any input, and storage chests as a buffer. All of these operate on the same network, and individual chests aren't really treated differently. To expand on that, storage chests really just form one big supply from which bots can take; same for passive provider chests. Active provider chests send to this supply or to any open requester chest. There aren't player decisions being made here, it's essentially just become a game of increasing and balancing throughput through hard numbers. In my opinion, the logistics system should resemble something like programming; items are values of certain types stored in variables (chests) which need to be delivered in a consistent and fast manner. I'm not sure as to how it would work in its entirety (others can add their ideas), but I think that making the logistics system more complicated falls in line with what the game is about in the first place, that being increasingly complex logistic problems solved by further and further automation.

It doesn't add any complexity because you don't care how well it is working, how much resources you're completely wasting, and how much power you're wasting. Which, to be fair, doesn't matter because the game is designed to make resources essentially meaningless and seemingly infinite. People only think bots are overpowered because the game fails to make obtaining resources meaningful enough to matter. This isn't the logistic bots fault players don't value the MASSIVE amounts of resources you're wasting to completely ignore every single challenge the bots have.

[Scindix]

In case anyone interested in my humble suggestions...
I came up with two ways to nerf bots very effectively, so that belts stay necessary:

• 1) Make some items "too heavy" for bots. After all they are small flying machines. How should they lift materials that weigh ~¼ metric tonnes? That way belts have to be used for certain materials like steel bars or crude oil barrels. Of course I wouldn't like the idea that these materials couldn't be delivered to the player as well. But maybe there is a workaround for that. Like multiple bots can team up to lift these things just in this special case. Or something like that...
  
  2) Another idea is that all bots on the whole map need a central processing unit for coordination and you can only build one. The number of bots this processor can manage is limited. You may upgrade it via research, but the cost in resources grows exponentially. If there are more bots than the processor can handle some of them might just stop working randomly so that there are never too much bots on their way (the more boring solution). Or they may start to confuse everything. E.g. delivering items to the wrong location, starting to crash into each other, dropping items randomly, flying into the wrong direction, etc.

Despite all that: Whatever you do make it optional. I would love new ways to play the game, but I also love playing the current one. And I don't want to be forced to downgrade just to reenable bots.

[Jürgen Erhard]

Notes:

"Belts are more fun". Who says? You will find that people generally gravitate to what is more fun and do it voluntarily. What is more fun, is what most people do.

"Bots replace the functions of belts". No they don't. On the whole bots replace the function of belts taking things over long distances, and things that are rarely used. Bots are pointless in small production areas as they do not transport thing with consistent speed as belts do.

And "more fun" is *SUBJECTIVE*! Not everyone thinks "main bus is more fun", even though most people play it. A lot of people play… what everyone plays. Look up "builds". Every MMO has them, WoW always has/had "seasonal" best builds. And given that you can inspect the equipment and skills of a char there, if you didn't have the (current) "best build" you very often didn't get to join a raid. Seasonal means it changed whenever the devs "nerfed" (generally, changed) something that made some other build "better".

And I hated that aspect. It takes away variety, especially in a "classic" MMO like WoW were your equipment is visible, because then everyone with this "best build" looks pretty much identical.

[Aardwolf]

I love the repair and construction bots and bots that fill player inventory.

I never use logistics bots to transport from/to buildings though, they're not really gameplay, they're boring. Except those that fill player inventory.

Best bots for me are repair bots from personal roboport, basically those and inventory filling bots are non game breaking bots that replace mouse clicks for you, which is good, convenient and awesome.

P.s. I also dislike combat bots due to their temporary lifespan. What's their point when my armor has lasers? Also, why is my armor better than a *tank*? Makes no sense. We should at least have laser tanks, or bots that fill tank ammo

[Kaiku]

Never liked mass use of logistics bot within the factory itself. They trivialize logistics and make Factorio not feel like Factorio anymore, or rather, it makes a factory not feel or look like a (typical) factory. I also like the challenge that belts provide. Making big blocks of assembly machines next to chests filled with random crap for the bots to effortlessly sort through is not challenging.

For what it's worth.

[stm]

I didn't read all the posts, so propably a lot has already been said but I wanted to give my opinion anyway:
I do not really like bots. I tried to build a bot based factory in my firtst playthrough after researching them in 0.14 (or 13?) and was less than impressed. They used much too much power for my taste (and I was going pure solar at that time!) and the throughput was terrible.
I also tried to build a buffered setup with bots which easily leads to bots moving Items from Chest A to B only to have an inserter move the Item back to A (That is why buffer chests are beeing implemented now)
For example I had one assembler in my refinery which produced processing modules which were used elsewhere a few screens away in my small production enclave. There normally were Items in the provider chest and available bots and still the requester chest was always empty.The transport was much too slow, since I obviously did not request enough overhead for the distance. I understand in 0.16 this is supposed to work better, which I did not test, but it shows one essential weakness of buts: They are not deterministic, at least not in the way you can control belts, where you easily can stear the items where you want them and only direct additional supply to other users and know when the items will reach their target.
So currently I use Logistc-Bots only very sparsely to transport Items with low throughput (e.g. trail fuel) where laying a belt would be incconvinient due to spatial constraints. The rest is done by inserters (offten without belts!), belts and chests.
I also like construction bots for blue-prints (especially since the range of the personal roboport is larger than the manual construction zone, but even there e.g. for floor tiles Bots are just too slow.
And for repairing defenses robots are actually neccessary in this game. Otherwise an autoheal function would have to be included or larger bases would not be possible since you would not be fast enough to repair all the walls and turrents. I hate the behaviour of the repair in my personal roboport though. It is too often, that if you fight near some structure one bot starts, gets damaged and destroid, only to have the next one to repeat the whole thing. Ther should be a possibility to better control them in this case (e.g. only repair if not currently under attack or something like it).
I also tried combat robots once and was verry disapointed. For the hassle to build them they hat much too less power and I had to start them myself and they were one-use. Better include them into the robo-ports and let them be crushed by biters (instead of construction bots which are trying to repair an orphaned wall segment). This at least could be balnced by HP etc.
So in conclusion: Even as somepone who does not really like the way bots are atm. Removing Logistic-Bots would be a huge mistake in my opinion, since belt controll ((un)ballancers, splitters, etc.) takes up too much space compared to assemblers in many cases and we are quite limited in the selection of inserters (e.g. there is no inverted filter inserter) so belt logistics sometimes ocupies more space then the corresponding assemblers.
Stm

[Ryba666]

In my opinion after reading this topic it is ok that bots are some OP becouse we have natural progression from yellow belts into red/blue to robots. It is same situation like change from stone furnaces to electric furnaces. Nobody cry about eletric furnaces is OP vs stone furnaces.

[Drury]

In my opinion after reading this topic it is ok that bots are some OP becouse we have natural progression from yellow belts into red/blue to robots. It is same situation like change from stone furnaces to electric furnaces. Nobody cry about eletric furnaces is OP vs stone furnaces.

Actually, you'd be surprised. Not that I'm one of those people, but there are definitely people who find electric furnaces too cheap.

[rldml]

Nobody cry about eletric furnaces is OP vs stone furnaces.

Because they aren't op. Which furnace fits best is a matter of investment cost of ressources, energy consumption and used space. Even the simple stone furnaces from the beginning of the game can be useful in late game, even though almost everone don't uses them at that point.

Greetings, Ronny

[jasendorf]

"I don't like bots, so they must be completely removed from the game or else I might use them against my wishes." What?

I'm about 95% belts, 5% bots... but where I use the bots, I want the bots. Not sure why they have to be removed... is there some competition out there where one person having bots wins and another not having bots loses? I guess since I play all single player, I don't know what the problem is.

[js1]

It should be noted that you can do this using belts -- you just need a closed loop, and then either a circuit network contraption, or just to re-gather all of your ingredients back up via filter inserters. Dump from chests into belts, merge belts together with splitters; go around the loop; filter back into chests. It has astonishingly low throughput though.

A miniaturized version of the train system -- preferably with logistics capability -- would be a really cool alternative.

Actually, what you mention in the first paragraph is good enough as an alternative. I am working on a small base around this concept, an improvement over a similar base that I did earlier (viewtopic.php?f=202&t=25107). And for scaling up, trains are great (see https://www.youtube.com/watch?v=DH3hCy4l4sY).

One thing that I find unfair is that bots use electricity, while trains require fuel. I think it would be nice to have either electric trains and/or higher tier logistics bots should use fuel as well (provided to control towers).