Hydraulic vs electric actuation in robotics | Marc Raibert and Lex Fridman

Dynamic control approaches that predict the future motion of a robot over several seconds ahead and pre-emptively adjust control signals produce more natural movement than reactive servo approaches that only correct based on the robot's current state.

“taking a dynamic approach where you think about what's the evolution of the motion of the thing going to be uh in the future and having a prediction of that that's used at the time that you're uh giving signals to it as opposed to it all being servoing which is servoing is sort of backward looking it says okay where am I now I'm gonna I'm gonna try and adjust for that but you really need to think about what's coming”

Humans, including most gymnasts and athletes, do not actually understand the physics of how they perform their movements; they learn through coaching and practice but typically lack explicit physical and mathematical understanding of the mechanics.

“unfortunately though when you humans don't really know how they do it yeah right we we're coached we we have ways of learning but do we really understand in a physical in a physics way uh what we're doing probably most gymnasts and athletes don't know”

Running appears more natural in robots than walking because running involves more aggressive leaping motions where the robot is more continuously in dynamic motion, whereas walking's intermittent contact creates control challenges.

“it just shows that the more aggressive and kind of the more you leap into the unknown the more natural it is I mean walking is kind of falling always right and something weird about the knee that you can kind of do this folding and unfolding and get it to work out”

To achieve sufficient rotation rate in 3D somersaults, robots must perform tucking motions (withdrawing the legs) to accelerate rotation, similar to how human gymnasts tuck to spin faster.

“in order to get enough rotation rate you needed to do tucking also uh you know withdraw the legs in order to accelerate it”

The basic design of hydraulic control valves had remained largely unchanged since the 1950s when they were designed for airplane safety, with little meaningful innovation despite potential for significant improvement through new circuit designs and integrated components.

“most of hydraulic uh control that is the valve that controls the flow of oil had been designed in the 50s for airplanes it had been made robust enough safe enough that you could count on it so that humans could fly in airplanes and very little Innovation had happened”

Compliance (spring-like elasticity) in robot leg design is important to how natural walking and running works, with proper spring behavior enabling efficient motion transfer and stability.

“there's compliances compliance me spring in the the design that are important to how it all works well”

For acrobatic maneuvers like somersaults where the robot must stick a landing, the control system must plan much further ahead than normal locomotion, pre-calculating momentum and rotation requirements at launch to ensure the landing is within reach.

“if you're doing something like a somersault you're looking out a lot further right if you want to stick the landing you have to get the you you have to at the time of launch have uh you know momentum and uhh rotation all those things coordinated so that a landing is Within Reach”

For typical robot locomotion like walking on even terrain, a limited prediction horizon of approximately 1-1.5 seconds is sufficient, with the control system recalculating at roughly 10 times per second, updating predictions based on current obstacles and foot placement.

“I think that the number is only a couple of seconds for spot so there's uh limited Horizon uh type approach where you're recalculating assuming what's going to happen in the next a second or second and a half and then you keep iterating you know at the next even though a tenth of a second later you'll say okay let's do that again and see what's happening and you're looking at what the obstacles are where the feet are going to be placed”

Boston Dynamics achieved the first 3D somersault with a 2-legged robot while at MIT, with graduate student Russ Tedrake who was also a champion gymnast before becoming an engineer.

“R Rob was my graduate student and we were at MIT which is when we made you know a two-legged robot do a 3D sumers salt for the first time um there we in order to get enough rotation rate you needed to do tucking also”

Boston Dynamics designed new hydraulic control valves with novel circuit designs that performed computing functions integrated into the valve itself, making hydraulic systems much more efficient, smaller, and lighter than previously possible.

“our Engineers designed valves uh the ones that are in uh in Atlas for instance that had new kinds of circuits they sort of did some of the Computing that could get you much more efficient use they were much smaller and lighter so the whole robot could be smaller and lighter”

Hardware innovation remains important and necessary for robotics; people who believe hardware innovation is no longer needed are mistaken.

“I think having good Hardware is part of the story and people who think you don't need to innovate Hardware anymore are wrong in my opinion”