Knee Joint Actuator Design

[stuckinthenavy]

I am designing a dynamic hexapod that has springs in the lower limbs of its legs. The overall concept is kind of detailed but I am stuck on an issue of joint actuation? The legs will have hip joints that just rotate through a limited range during the stride and recovery phase of each leg. The problem is when the leg is swinging back forward again, the robot chassis will be lower a little because of the springs in the legs and the recovering legs will not have the ground clearance to come back to the front. I need to design a knee joint that bends the lower half of the leg up while the leg is swinging forward again and holds the leg straight during the stride portion. I want to add some elastic bands to the back of the joint that would prevent it from hyper-extending forward as well as providing some cushion to the joint. I have a picture of the concept if the writing is unclear and I will post it if anyone wants to see it.

The joint needs to act quickly but will have no real opposing forces when acting. I was thinking possibly putting a servo in the center of the body and running thin cables down each leg and out at the "hamstring" and connecting to the "calf" so when the servo twists, the joint would contract, and then when it rotates back, the joint would extend due to the elastic forces from the elastic bands on the joint. But I don't think this would work fast enough and the actuation need to be of variable speeds.

Subnote. Does anyone want to work on this project with me? I would love to talk it out with anyone interested. Overall concept is a hexapod that has spring like legs and can run. Dynamic motion control system, as well as having an off balanced rotating disc that rotates in phase with the stride to provide additional momentum and store extra potential energy in the springs in the legs.

[stuckinthenavy]

What about some complex locking mechanism so the acutator isn't under load during the stride? anyone else see this in their head? Maybe we could sketch out a few concepts.

[Adrenalynn]

Hi, and welcome to the TRC. Just a heads-up - I expect replies will be slow for your interesting concept for the next few days. A large chunk of the TRC is off at Robogames in San Francisco for the three day weekend event. I'll be heading down there myself tomorrow morning...

[stuckinthenavy]

Thanks for the heads up! I was curious why all robotic forums kind of died out. I'll sit here and wait paitentily for every one to come home.

[Larry]

While you are waiting, you might want to google "passive dynamic walking" (particularly articles by Andy Ruina). Links should appear for biped walkers with knees that lock extended and unlock with a small solenoid during the recovery portion of the gait. The only "actuator" is the mass of the leg making the knee bend. Seems to work. I

[Adrenalynn]

Great search strategy, Larry, thanks! Sometimes it's all about knowing/figuring out what to search for.

Stuck' - [thank you for your service!] - largest robotics event in the world (according to Guinness) http://www.robogames.net

[stuckinthenavy]

Thanks for the search tip. I have been looking into similar knees as the runbot. If I put a pressure sensor in the "foot" could I have two states, touching the ground and recovery? How would I work out a timing system? Should I try to design this to run at one speed or should it be an adaptable program that adjusts all the stride and recovery actuators based on speed over ground?

[stuckinthenavy]

I think I want to do a knee joint similar to runbot with a servo in each joint but I was curious how I go about deciding which servo I need to suit my needs. I want one that will be hackable but I don't want one that is too big. What parameters do I use to start picking the servo that is right for me?

[Adrenalynn]

Three biggies:

Torque required
Torque required
and Torque required.

After you answer those three, consider the torque that you're going to require.

Beyond that it's easy. ;)