Pacific Rim Control

Understand: even if you have a mechanical system capable of perfectly reading a human's movements and translating that to robotic limbs, the mere act of swinging your arm while strapped to a bunch of mechanical sensors / robot arms does not necessarily provide ALL the information necessary for a fluid, human punch.

If there's only one sensor on the fist, the robot can only know the fist's orientation. It has no idea what the rest of the arm is doing.

Put additional sensors on the wrist and forearm and you still don't know what the shoulder and chest are up to. You've got to strap sensors all the way down the line.

Even then, the robot now only knows how the limb is positioned and its acceleration. It has no idea what the individual muscles of your pilot's arm or body are doing; it's using its own computer power to determine how best to go about moving its own limb to result in the closest human approximation, not EXACTLY the way the human is.

Robotic hydraulics are not the same as human muscles. The way a robot goes about walking is fundamentally different from the way a human does. Our feet are pendulums that swing forward and rely on gravity; a much more enormous robot whose weight is off for its scale will not walk 1:1 like a man. This pendulum action is the reason astronauts hop on the moon instead of walking around.

Taking this further, what about the water that the Jaegers are walking around in? Are you going to flood the cockpit with a viscous solution to simulate that difficult terrain?

So let's assume you've got some high-tech bodysuit that knows the position of every limb of your pilot and is translating that across to the robot. It's still not as perfect as knowing what the muscles are doing in addition to the above, but it's the best you've got. If you want real robattlin' expertise, you need some degree of force feedback so the pilots know what's going on, so build that in to the suit as well.

But what if the robot gets knocked down? Your humans are strapped in and still standing. You don't want them flailing around in the cockpit making unnecessary movements that won't help the robot as it flounders. What if the robot's not flat on its back, but leaned against a building; how can the humans "push off" of that when there's nothing in the cockpit in the same position for them relative to the robot and the building? The environment that the human pilots have to interact with greatly differs from the environment of the robot.

It should be clear by now that for the most precise movement, you need more than just a motion suit. Those stabilizer arms are a step in the right direction. They can keep the pilots at a relative angle to match the robot when it's tilted in such a way that someone standing in side would have no chance but to try to stand up straight or fall over. Instead of being used just to read the pilot's movements, they can be used to apply some of that force feedback. You can't look at this whole system as only transmitting information from the pilot to the robot; it's going in both directions. That's how humans move. Nerves tell the brain what's going on, it reads balance information from the ear and knows, to an extent, where its limbs are, and then sends movement information to the muscles which act on it and constantly update the brain on their progress, so when you step on a Lego you don't put all your weight on it.

But you can't get all that sophistication just from a force feedback suit and some mechanical arms jostling you around. If you want to pilot the robot like a second skin, know what it knows, have it move like you move.. you're going to need a NEURAL INTERFACE. Now sensory information from your robot goes straight to your pilot. It can hear an approximation of what the robot hears, see an approximation of what it sees, and move much more realistically.

When something goes wrong with the movement, for example, the robot slips, what's the reaction time between a neural link and purely mechanical piloting? The brain can react as quickly as the information processes and moves around, while under a mechanical system, you have the additional step of waiting for the robot's problems to become apparent to the pilot, THEN his processing, THEN his corrections. The neural pilot is working at the same time the robot's computer is.

These tiny little reaction issues don't just come up when something goes wrong with a movement, but they're needed throughout the whole process to properly maintain balance. The human brain is constantly running away in the background keeping you upright as you walk or run, making adjustments in such a way that you aren't entirely conscious of them.

And, again, the robot body is vastly different from the human body. It's weighted differently. Its muscles work differently. If you were to put the same stresses (scaled down, of course) on the human body to represent the robot, they'd likely be flailing all over the place. More so for Jaegers that do not have the usual human form and proportions. The computer's already pounding away and crunching all this data, making the best translations from robot sensory information to human physiology to human movement to robot movement as it can, but you can always "do better".

The best computer we have is the human mind. It intrinsically understands movement in a way that we can't yet code computers and robots to understand. If it's feasible to hook these things up and get them to talk, why not do it? Pacific Rim posits a universe where we can, so go for it. And why stop at one mind? Dual core those human brains. Get some hyperthreading up in this.

TLDR: Looks cool. Sci-Fi.