Bio-mechanics in Power armor design and hybrid weapons platforms.

Note. This applies to everything, not just Mass Effect. Also, sorry if it’s kinda rambly. I put it together in pieces while waiting for battletech to load.

The issue with unusually shaped piloted weapons platforms is ease of use. Not from a strength standpoint but on a balance and sense of positioning one. Input ease and lag are issues, and speed is also a consideration though only really becomes and issue with heavier platforms.

Imagine you’re piloting a mech. If the weight’s off, you’ll automatically have trouble piloting in a natural manner. Sure compensators can handle basic balance issues, but for smooth movement you’d need to cut them out and learn to compensate via your own sense of balance. Otherwise you’ll be fighting the mech for control instead of fighting the enemy shooting at you. By keeping the center of balance in the same area a human’s is by weight management, you eliminate that issue and enable the pilot to move in a more natural and free manner.
As for sense of positioning, I mean like how you generally know where your arms and legs are without needing to look for them. Thus, if you’re piloting a platform with inhuman proportions or odd protuberances, it would screw with your body image. Think on it like suddenly having proportions like a gorilla, or things sticking out of you (IE weapons, sensor booms, and so on). You’d need to adapt your sense of positioning every time you got in a mech, and if it was a nonstandard configuration, it would just be that much more difficult. That would also slow you down and make movement more clumsy. By keeping a strictly human proportioned form, you solve those issues and would only need to deal with the size differences. That also makes keeping mechs as small as possible, or sticking to powered armor, extremely desirable.
Also, as a side note, jump jets would need to be located in places the pilot species find as ‘natural’ as possible. Otherwise the pilot would suffer a massive body positioning issue and require extensive stabilization and/or training to learn to compensate so they don’t pull a team rocket. General locations would be hips, lower back, feet and/or lower legs. Oddly placed thrusters could be gotten used to with extensive training however.
Input ease and lag are the next major issues. IE, how fast/precisely does the operator’s intent turn into action by the suit/mech. There are several potential solutions. Motion sensing, IE, sensors in the cockpit detecting when and how the pilot moves. It’s got the greatest lag and lowest precision since it has the greatest detachment. Second, and arguably most practicable, is muscle nerve monitoring. Instead of the suit sensing the body moving it senses the body wanting to move, much like some kinds of prosthetics. Combined with some form of force feedback, this would make for a solid ‘Functional’ control system for most people. Last would be two levels of cybernetic link. The first is a direct splice into the central nervous system to do the same as muscle nerve monitoring but for the entire body at once as opposed to group by group. Requires pilots being cybered (A datajack or something like the Alaya Vijvana from Gundam) with an interface however that would make them susceptible to electrical attacks and potential system hacks. The other is full body virtualization. IE to shut down the pilot’s body except for essential functions, and merge their nervous system with the mech/armor. the issue would be the same as previous but even more so, plus when piloting it would make even mild electrical surges potentially lethal and any ‘shock’ the mech would feel would echo in the pilot’s mind. Both have high speed and accuracy, but require dangerous surgeries to
Complex systems on mecha, such as multi-arc weapons, detailed EWS, and so on are a huge hell no as they’d overload the pilot’s mind. Think on it. You’re driving your car, la de dah, and then someone rams an AWAC into your skull. You’d crash under sensory overload, and that’s not counting having to split your focus between piloting and the rest. Making a mech big enough to incorporate a second pilot to handle that is possible, but makes the mech itself larger and harder to pilot. Remote or AI operation of said systems is the most realistic option.

Powered armor (Like the Iron Man suit) are a whole other mess. First you’re working with minimal space. Second, balance is even more of an issue since the suit’s too small to incorporate a full stabilization system. Three, matching the flexibility of a human without it being impeded by the suit is also an issue.
Minimal space is going to limit your systems options, though bulking up the suit’s certainly possible. It’d slow down it’s reaction times and make it harder to pilot however. That means you need to strictly ration space for weapons, jets or thrusters of some kind, and sensors, not to mention the essentials like power and basic motive systems. That means sacrificing this to get that, or minimizing that to get more of this. You can bulk up a suit, kind of like the MADOX or Starcraft Marine armor, but that starts biting you in the ass fast because the bulkier it gets, the harder it is to control and the slower it is. Diminishing returns, essentially, until you have more problems than you solve and you’d be better off going to a hybrid platform like an Appleseed Landmate, or a full fledged mech.
Flexibility is just as much if not more of an issue as it dictates size and shape limitations. The armor has to flex around the body in a way that doesn’t encumber or crush it. That’s going to mandate certain size limits, especially in the groin. Do the proportions and/or movements wrong and you can rip the pilot’s legs off.
On the plus side, powered armor can use modified personal scale or light vehicular scale weapons so converting pre-existing stockpiles to be power armor usable would be easy.