note that legged animals (like humans) have self-tuning (biodynamic) spring-damper systems built into our legs (and backs). if we didn’t, we’d not have the range and endurance that we enjoy, and possibly would have died out as a species due to an inability to find enough food fast enough. nature is amazing.
the sloths would be fine but not necessarily us. a differently-evolved monkey (maybe chimp or gibbon offshoots) might be, but likely not humans, as nature specifically selected us, from a biomechanical perspective, for efficient walking/jogging. we're not evovled to sit and stand around, as we're seeing very plainly in modern health outcomes.
Tuneable stiffness is a basic property of muscles. You can think of a muscle as a spring-damper combination for which you can set the spring constant, zero point, and damping factor. That's what allows elastic energy storage in running. In human runners, you get maybe 70% of the energy back on each stride. Cheetahs get 90%.
Building an actuator with those properties is tough. There are a few approaches.
One of the best is a double-ended pneumatic cylinder. If you can control the air pressure on both sides of the piston, you have a spring with adjustable stiffness. This can store energy. Raibert's original hopper used something like that for the hopping motion. You need fast proportional valves near the air cylinder to make this work. Also, compressed air. Festo, which builds advanced pneumatic systems, does this sort of thing.
This can be done with two opposed springs, each with its own drive motor. That's been done a number of times. Tighten up both, get a stiff spring. Loosen up both, get a limp spring. The mechanical setup tends to be bulky and involve pulleys and cables. CWRU has built robots that way.
There's a way to fake it using something called a "series elastic actuator". This is a single screw drive linear actuator with a stiff spring on the end and a sensor to tell how much the spring is compressed. If a force is exerted on the output end, the screw drive motor cranks fast to try to stay ahead of the applied force and create the illusion of a spring with the desired stiffness. There's no energy recovery. In fact, as a force is applied, not only do you not store energy, you expend energy to move the actuator. So this is mostly for research projects and things that don't run on batteries.
There have been various efforts to make muscle-like actuators, using fluids that change properties as electric or magnetic fields are applied. I don't think any of those devices became useful.
Linear motors can do this directly, but linear motors tend to cost too much, and have low power to weight ratios. With some designs, you can do regenerative braking and get energy back. There's never been a killer application for linear motors, and so they tend to be small volume items.
It's one of the mechanical problems in robotics with no really good solution yet.
6 comments
[ 4.0 ms ] story [ 27.0 ms ] threadCould someone with a university account GET this and post it somewhere please? Seems very interesting.
https://mega.nz/file/oB8z1aoJ#nm3pz9LO4uGMwMiQMh6SbO3PZmfTnR...
Building an actuator with those properties is tough. There are a few approaches.
One of the best is a double-ended pneumatic cylinder. If you can control the air pressure on both sides of the piston, you have a spring with adjustable stiffness. This can store energy. Raibert's original hopper used something like that for the hopping motion. You need fast proportional valves near the air cylinder to make this work. Also, compressed air. Festo, which builds advanced pneumatic systems, does this sort of thing.
This can be done with two opposed springs, each with its own drive motor. That's been done a number of times. Tighten up both, get a stiff spring. Loosen up both, get a limp spring. The mechanical setup tends to be bulky and involve pulleys and cables. CWRU has built robots that way.
There's a way to fake it using something called a "series elastic actuator". This is a single screw drive linear actuator with a stiff spring on the end and a sensor to tell how much the spring is compressed. If a force is exerted on the output end, the screw drive motor cranks fast to try to stay ahead of the applied force and create the illusion of a spring with the desired stiffness. There's no energy recovery. In fact, as a force is applied, not only do you not store energy, you expend energy to move the actuator. So this is mostly for research projects and things that don't run on batteries.
There have been various efforts to make muscle-like actuators, using fluids that change properties as electric or magnetic fields are applied. I don't think any of those devices became useful.
Linear motors can do this directly, but linear motors tend to cost too much, and have low power to weight ratios. With some designs, you can do regenerative braking and get energy back. There's never been a killer application for linear motors, and so they tend to be small volume items.
It's one of the mechanical problems in robotics with no really good solution yet.