How many DOF should mech arms have?

My brother and I have been playing around a bit with mechs in VR. I currently have a prototype where you can sit in the cockpit of an AMP suit (from Avatar, see below) and as you wave your hands around inside the cockpit, your make waves its giant arms around outside.



To me, this is how mechs would most likely work; you have hand controls inside that allow you to directly control the mech arms. (Legs are much harder to directly control, so I'm fine driving those around with joysticks, with some AI running on the system to safely place the feet).

But at this point I run into doubts. Some mechs are more humanoid, with high-DOF arms that it makes sense to aim independently:



...but others are less humanoid, and often have stubby or fixed arms, which you would aim more like a tank:



And of course others are in between. I realize that variety is key to the appeal of mechs. But this seems like a very fundamental design decision, especially for a VR game. Do your arm guns have high DOF which you control directly by positioning your hands? Or do they have low DOF, and you aim them by rotating your torso and pitching the guns up and down, like a tank turret?

If low DOF, you may as well be playing with a gamepad on a flat screen. But with high DOF, it basically becomes like any other first-person shooter, except you're 15 meters tall. Which is more distinctive and fun?

Hoping to hear from fans of mech games in particular. My last bout of mech mania was some years ago, though it's starting to come back.

 

In oldschool tabletop Battletech rules the degrees of freedom that the arms have mainly come into play in melee attacks. It's harder to hit the less joints an arm has. Firing arcs play a big role too, weapons mounted on the torso cover an arc of the 3 hexes counting from the base the mech stands on, arm weapons cover the same plus one more hex to the side where the arm is.
There is one special case described in Battletech lore and I think also in the rules, though I'm not sure if it's part of the standard ruleset. Mechs that have just a shoulder joint in the arms can flip them 180 degrees so that they aim behind them. Mechs like the Rifleman have that ability:



I think in your case you'll probably need to think less about degrees of freedom than about coverable arcs and the movement speed with which targets can be acquired. Aiming an arm mounted weapon should be faster than rotating the torso, a light mech should be faster than a heavy mech.

I'm trying to remember what Battletech lore says about how the controls work and I have the vaguest of memories that the mechs that have both hands and arm mounted weapons have different controls for whether you want to aim and fire or use the hand to pick something up. The rules cover picking up trees to use as improvised clubs for example.


What is your target plattform VR device? The Oculus Quest? I think a lot of choices you might need to make based on hardware constraints. E.g. flipping weapons to your back probably makes zero sense as a feature in such a game and would on the implementation side require a screen in the cockpit that assists with aiming backwards and you might not have the performance budget for rendertextures.

My suggestion would be to focus on what fun things you can do with the cockpit itself. I remember that Mechwarrior 2 had a quite extensive list of hotkeys and playing that game keyboard only had you frantically pressing all kinds of buttons. I thought that was quite immersive in its own way. There are mechanics around heat in those games, where you build up heat from shooting, and the reactor will auto-shutdown at critical heat levels, leaving you immobile. You can slam an override button to override the shutdown, but if your heat level gets too high your ammo can explode or the pilot can pass out. I'd be tempted to make the user interact more intricately with virtual buttons in the cockpit. E.g. instead of making the reactor auto-restart, make the user press a series of buttons, and also have that bootup sequence at the start of a mission to make sure they know what to do when they'll need it. Of course that would go counter to the casual nature of VR games on these phone-like VR systems. Maybe an arcade mode and an immersive mode would make sense? Look at me feature-creeping your game, I think I'll better shut up now...

Just one more thing: The reactions I've seen from (non-vr) players to devs trying to make controls anything other than what they are used to, even if there are very good gameplay reasons for it and the new control scheme offers a higher skill ceiling and generally makes more sense withing the mechanics of the game... were almost universally bad. So I'd first look at how the established vr shooters and possibly mech games handle it, and then you're probably best off copying it if it is a good implementation that is widely accepted as gold standard :-/.

 

Thanks Martin! There's some good feedback there.

I think there isn't a universal standard for mech VR games yet, and I think I've observed that VR players are generally eager to take advantage of the VR input options rather than copy flatscreen controls. So I think there's an opportunity here to innovate that might not go over like a lead balloon.

In fact I just learned there was a PC VR game (Archangel) which had somewhat similar arm controls to my idea. The one person I've found who played it didn't like that game much just because it was too arcady; they were looking for a deeper mech-piloting experience, with lots of systems to manage. I tend to agree and it sounds like you do too.

One interesting thing I've learned this week, in playing around with different mech models in VR, is that most of them have really poor visibility from the cockpit. In most cases you can't see the ground anywhere near you, so a smaller mech could be casually dismantling your legs and you might not even know! Of course there are designs with better visibility, and you can always supplement windows with sensors. And in a real battle you're probably not standing around in one place long enough for someone to sneak up on you.

As for the arms, I 'm still torn. Switching between a "weapons mode" and a "hands mode" isn't a bad idea though. In weapons mode, the forearm guns aim at whatever your hand is pointed at, so if done right, aiming should feel very much like having a gun in each hand. In hands mode, the wrists are unlocked and we do full IK on the position and orientation of the hands, so you can actually grab stuff. And then of course we also need a mode where you've let go of the arm control entirely, so you can punch buttons within the cockpit. Complex? ...We'll just call it "deep."

 

What hardware are you targeting and what tracking capabilities do its controllers have? How many buttons and analog sticks are there on each controller?

Have you ever played Terranova Strike Force Centauri? It was way ahead of its time imho. Just googled a random youtube video of it:

And a newer title - MWO:

I think Mechwarrior Online had its own thread here some years ago, where it was called "the thinking man's shooter". I thought it did some things quite well in regards to setting itself apart from regular FPS games, but it didn't manage to pull me in. The video I linked mentions how they implemented gameplay differences between arm mounted and torso mounted weapons I think.

Then there's Hawken, which felt too much like a fast paced shooter imho:

Have you looked at this VR title yet?

https://store.steampowered.com/app/334540/Vox_Machinae/

 

Targeting Oculus Quest. It has two 6DOF controllers, each with a thumb stick, two thumb buttons, a trigger button, and a grip. And one with a menu button (the other menu button is reserved for the system.) So, plenty of inputs really.

Thanks for all the reference videos — I'm going to pop some popcorn and study those tonight.

 

Enjoy! The first two are quite entertaining imho. The third one I suggest to just skip through, watching the whole thing won't tell you more, it's just frantic gameplay without commentary. The fourth one probably is most relevant to the problems you are trying to solve.

Nice! That should give you a lot to work with.