Robotics Game Idea only have document

You can post your idea wherever you like; here is fine. Be advised that you'll likely only get advice on how good or not good your idea is. Not many people are looking to implement someone else's idea, because generally speaking, modestly experienced game devs have their own ideas. Plenty of 'em, more than they can get done. Ideas by themselves aren't that valuable. It's turning them into actual games, that's valuable.

If you want to get your idea turned into a game, likely, you'll have to learn some aspect of game development that helps you do that. Like programming for instance. Either that or you'd have to become very very good at persuading and leading other people. Definitely not you at this time, as if you were "all that" with amazing people skills, you wouldn't even be asking what to do next. You'd be hustling your project.

Thank you for your response! I think I'm going to put my idea on the back burner for now. But for when I do decide to try developing it, what might your recommendation be for software/programming that would allow a beginner to make a 3D physics game with realistic graphics? The goal of my game is to teach people how to build things with real robots/engineering skills.

Lordy. Accurate physics is a lot more of a requirement than just making a game. It's a decent idea for someone to pursue, but without the hardcore programmer knowledge, I don't see any way for you personally to pull it off.

If you have the physics knowledge already and it's only a lack of programming skill holding you back, then possibly you could find a physics package that is "deemed to be accurate" by your physics peers, and then cobble some rudimentary programming stuff on top of it. But someone else will have already had to have made that accurate physics package. I don't personally know what that package would be. Although I did intend to do physics and holography when I first entered college, circumstances caused me to rapidly move away from that.

As for "realistic graphics", I see that as a non-goal. Your idea lives or dies on the accuracy of physics simulation, if you're trying to have it be applicable to real robots. Even clunky 3D models could have some slightly nice illumination and surface rendering done on them. Probably won't win any art direction awards, not unless some of the excesses of the Fine Art world catch up with gaming, and we get bananas tacked to walls or some such.

Ris.Lee wrote:

what might your recommendation be for software/programming that would allow a beginner to make a 3D physics game with realistic graphics?

Any game engine (Unity, Unreal, Godot...) supports rigid body physics simulation. Those can all do something like Half Life 2.

If your ambitions regarding true robotics are higher, i can recommend the Newton Physics Engine (open source C++ on github). It's still a physics engine for games (e.g. Penumbra, Amnesia, Soma), but with much higher accuracy. I use it for self balancing / walking ragdolls, and any other physics engine i have ever tried failed on that due to lack of accuracy.

However, real robotics is not for beginners. You need expertise in programming, physics simulation, control theory, inverse kinematics.
So it depends a lot on your game dev experience and on your games simulation requirements. If you tell more about that, i can tell if standard game engine should be good enough, which is more than enough to learn for a beginner.

Thank you! My idea is that the player will build robots. They'll build it piece by piece, more or less. The body they'll build love by piece, complex parts like engines or motors will be either simplified or provided as its own piece (I don't want one design to take too long to build). I'd really like it if the designs could be exported and usable in real life, too.

Might'be had some typos. Not sure how to edit after posting.

Are you familiar with the material at https://en.wikipedia.org/wiki/Robot_combat ? There's quite a lot to real world robot competitions, not the least of which are weight classes and safety.

Ris.Lee wrote:

what might your recommendation be for software/programming that would allow a beginner to make a 3D physics game with realistic graphics? The goal of my game is to teach people how to build things with real robots/engineering skills.

Figure out what you know you can do, and what you know you can't do. That's the gap you've got to cross.

A good game engine does a lot of the work for you. You'll still need to provide the game logic and the various assets, but the world containers, the graphics, the audio, the input handling, and similar is built in to the engine. At that point as a beginner, implement whatever parts you think you can do, and struggle over the parts you can't do until you figure them out.

Beginners especially don't know what's hard and what isn't. Often people spend weeks agonizing over what a story element ought to be, when in actuality, the hardest questions will be along the line of "how do I make a door do what I want?"

Thanks so much for all the advice! I'll definitely check out that Wikipedia page. This is going to be a project I tackle once I've got some experience under my belt. Until then, I'll keep researching and learning.

Ris.Lee wrote:

Thank you! My idea is that the player will build robots. They'll build it piece by piece, more or less. The body they'll build love by piece, complex parts like engines or motors will be either simplified or provided as its own piece (I don't want one design to take too long to build). I'd really like it if the designs could be exported and usable in real life, too.

Something like Garrys Mod, or games where players can build various vehicles, i guess.
For that there should be no need for specialized physics engines. I was just confused by the robotics term.
I would say, for mounted industrial robots, or for moving robots on wheels, standard game engines work well.
Only for robots on legs as shown by Boston Dynamics for example you would run into limitations.

Still, for a 'first game' (if so) it's quite difficult imo. Start with something smaller, but keep the idea in mind. ; )

JoeJ wrote:

I would say, for mounted industrial robots, or for moving robots on wheels, standard game engines work well.

Do your expectations change for realistic robot combat? I'd expect usual simulators to fall flat on their faces, as far as providing results that have real world verifiability.

bvanevery wrote:

Do your expectations change for realistic robot combat? I'd expect usual simulators to fall flat on their faces, as far as providing results that have real world verifiability.

Thinking of TV shows i have seen about little self made machines trying to destroy each other, i would not worry. Any physics engine should be able to simulate that well enough for such game.
Those are very simple robots, usually built wide but not high, with large contact support area and a low center of mass. So they won't fall over easily, and simulation error can't affect their balance. It does not matter if some grabber mounted on a tank only reacts with some latency, or if it overshoots the target a bit. It will still hit the enemy, and the tank itself will stay on the ground.

It would become very difficult only with humanoid robots standing on legs, where com is high and support area is small. Punching an enemy with the fist requires to counteract the swinging arms momentum by moving more pressure to the heels for example. If you get this wrong the robot already falls over just from moving its own arm. High precision is needed. The physics engine must respect laws of physics precisely, otherwise you can't predict the future and you'll fail even you could solve the difficult control problem correctly.
That's where game physics usually fail. They use hacks and approximations, prioritizing performance over accuracy. They can resolve the collisions from the combat, they can do breaking joints from heavy impacts, but their motors are too laggy and their joints are too soft to allow the precise control needed to keep a humanoid robot in balance.

That's where i would draw the line. Tanks and vehicles yes, Boston Dynamics Atlus no.

Still, various physics engines have different strengths and weaknesses, and comparing them is hard. Picking one engine over another makes still a big difference, i guess.
For a test, i would try to create a simple grabber like in an industrial assembly line. Some robot arm with 3 joints maybe, and it should be able to grab a box, picking it up, and putting it down at some target.
That's already pretty challenging both for the programmer and the simulator. If the engine of choice fails at it, i would try another to see how it goes there. Only after this works i would make my choice of engine final for further work on the game.

There is also a difference on what kinds of joints or motors engines support or not, which is something we can look up before considering some engine in detail. If engine X does not support motors at all, i can rule it out before wasting time on learning the engine.

JoeJ wrote:

It will still hit the enemy, and the tank itself will stay on the ground.

I don't think you're thinking about the destruction phase of a competition. Sure a game simulator can handle trading blows, between squat vehicles that are intact. But what about when vehicles start getting split open into funny shapes, and wheels get knocked and bent out of axis? Can the engine even handle physical deformations in the 1st place? I seriously doubt it; sounds "off-label" for typical game simulation. I just happened to watch a demolition derby for the 1st time in real life. There's a lot of looks like nuthin's happenin' until finally something does happen, for reasons that aren't always clear.

bvanevery wrote:

But what about when vehicles start getting split open into funny shapes, and wheels get knocked and bent out of axis?

Yeah, this needs some work. For destruction physics engines usually support convex decomposition to precompute pieces, but this won't give you sharp and precise cuts from a saw weapon for example. Bent axis can be done with a custom joint. Though, physics engines rarely allow to make your own joints easily. Newton is the the only exception here i know in this regard.
Imo, destruction in games is generally still more fake than realistic. But mostly that's fine, because destruction rarely is a game play mechanic but rather cosmetical.

bvanevery wrote:

Can the engine even handle physical deformations in the 1st place?

Some engines can do this using soft bodies. But i have not yet seen this for metals which then keep a deformed shape.
So yeah, this can be a reason to make your own custom engine or extending an existing one.
Which is a good thing. I'd like to see more custom tech and innovation. Currently all we get is Epic proudly announcing new engine features, which then every dev gets at the same time, so the potential advantage of having something new becomes just marketing hype and a new state of the art, but no exciting feature of a unique game to stand out.
Off the shelf engines with standard feature sets converging towards a common denominator are a big mistake for a creative industry imo. But they rarely listen to me. ; )

bvanevery wrote:

There's a lot of looks like nuthin's happenin' until finally something does happen, for reasons that aren't always clear.

That's why simulations are so interesting, because you get those unexpected things from that as well. To me it means i loose a lot of control over my game. But i also get emergent / unexpected stuff a random number generator could never do.

Anyway, you seem a bit doubtful regarding physics engines. Which is understandable, because the impression often is 'it just barely works, but still has many issues and lacks features.' That's true. But on the other hand i'm sure there are many unexplored options of what such simulations for games. I see most of the failures on the design side, not willing or confident to explore those options. Modern games have less rigid bodies and ragdolls than games of the 2010s, so i notice a decline rather than progress. That's sad and just wrong. And now, where further improvements on gfx become unafordable, maybe we should change our mind.

Bridge simulators were a genre for awhile, don't know about now. I suppose it's a physics problem for which reasonable composition, failure, and destruction results are possible. In something I judged for the IGF a long time ago, I would always make these super precarious bridges that were just on the verge of collapse anyways! My goal was to have the cars go across it the first time, to be lulled into a false sense of security. Then if it actually failed on the 2nd time I regarded it as a perfect trap bridge.

Thank you, everyone! I've gotten some really valuable insight. And I'm glad my game idea sparked some good discussion. Imagine what the final product will do for the gaming community! I'm excited for it.