I really like your designs. Have you considered removing the knee and using the linear motion directly to change the hip-ankle length? The may reduce your maximum torque requirements by removing the bent knee geometry.
Do you mean something like a telescopic leg? I think that would be extremely strong but very slow too (at least with the motors I am using), and in a quadruped you need to be fast. Note that the shank crank is 26mm, and the shank is 260 mm, then we are multiplying speed by 10 (and dividing force by 10 too). Maybe with a motor with a very high Kv (1700 Kv for instance) what you propose could be feasible... I will give it a thought!
Yes,I was thinking telescopic. You can get different ratios of rotation to screw-length which might help, although I'm not sure of the weight/size trade-offs.
You make amazing content. Thank you. I wouldn't use the same approach. Ball-screw mechanism takes too much space and heavy. I would use ballscrew only on one joint, and would use harmonic/cycloidal reduction on other two joints.
That is the approach used in Boston Dynamics Spot, and it seems to be a good one. I agree ballscrews take much space, however I think they are not heavy at all (200 g for 100 Nm). Thanks!
It is a very nice idea to use ball screws in all three joints but do you thing will you be able to fit this design in a quadruped chassis? It looks very big and not soo compact.
Yes they are a bit long, but they can fit in a 0.8 - 1.0 meter quadruped. I am trying to shorten them by avoiding the motor shaft coupling. Yo can see this in my next video. Thanks!