Having the legs in this configuration will improve performance but
doesnt fit the design i was going for… once again the design favor’s function over aesthetic!
I have learned it takes many revisions before you get it right. Even when you get it right, the feeling that it can be improved will always haunt you. When I built my sensor board, I made 5 prototypes, each one an improvement from the last. Whenever you scratch build something, there are lots of things you discover that require changing, and that’s what engineering is all about.
Your design is very interesting and the fact that you plan to skin it will make it very cool. I enjoy reading about your progress, keep up the good work.
-----------------------------[BODY/LEG DIMENSIONS]
Phoenix setup
CoxaLength con 29 Length of the Coxa [mm]
FemurLength con 76 Length of the Femur [mm]
TibiaLength con 106 Lenght of the Tibia [mm]
Stalker setup#1
CoxaLength con 60 Length of the Coxa [mm] 51% bigger
FemurLength con 130 Length of the Femur [mm] 41% bigger
TibiaLength con 140 Lenght of the Tibia [mm] 24% bigger
Stalker setup#2
CoxaLength con 60 Length of the Coxa [mm] 51% bigger
FemurLength con 72 Length of the Femur [mm] 5.5% smaller
TibiaLength con 124 Lenght of the Tibia [mm] 14.5% bigger
with set up #2 the size of the leg works out to be a bit smaller than the phoenix as well as the original stalker leg design#1.
i feel with this design the quad will be stronger but it will be lacking in performance as the gait step will be too short.
the smaller design doesnt look as good but does solve my problem with the torque!
here i have added the rest of the Tibia structure.
now i can complete the measurements and angles for the programming and also finish work on the vacuum parts… that i promis are coming soon.
Same here. im writing the cobe blind, as i dont have all the servos so i cannot test it but im happy with what i have so far so fingers crossed.
thanks.
thank you for your comments. i hope to get an update on vacuum parts soon. just had my second baby boy this week so will be busy with other things but asaic (as soon as i can) ill get it posted.
Everytime I check this topic things are looking even better then before! You’re getting close to the original picture. I know your not building a exact copy but the overall design is really close.
The bend tubes look really neat. How does that come together with the vacuum parts? I hope you’re still planning to add them. I hope the 645’s you’re planning to use can hold the long legs. When do they arrive?
I don’t have a quad on my desk to give you a hand but I’m happy to help you out when you’ve got questions. Would be fun to build a quad myself one day and see how my code holds up.
Your new curved tibias looks great! Awesome. Always a pleasure to see the progress of your project!
I’ve mentioned this before but unless you’ve not solved the kinematics for the 45 deg tilted coxa, that will be your major problem (beside the walking gait algorithm).
[/quote]
I’ve been thinking of a way to solve this and I believe I’ve found a solution. I’m soon going out for a long business meeting but I’ll try to shortly explain it:
The solution is to use rotation kinematics to convert the global body coordinates into to the local leg coordinates (for each leg). Since the orientation of the coxa are in a fixed position that makes it much easier to simplify the rotation matrix formula (the one I used for BodyIK in Xan’s code).
I’ll have to go, but I’ll try to explain more later.
I’ll try to explain more how to solve this. Since I have a bunch of thoughts in my head I’m not sure where to start…
First, imagine that your stalker is built with the coxa parts horisontal and inline with the body. The body dimensions in Xan’s code (coxa, femur and tibia length, offset values and coxa_angle=0) must be set using the construction configuration I just mentioned. In this configuration the legs correspond in the same coordinate system as for the body, meaning that the XZY axis for the legs have the same directions as the XZY axis for the body.
So, when tilting the coxa downwards about 45 deg like you did on the Stalker the two coordinate systems does not correspond anymore. The idea I have to solve this is to convert the real body coordinate values for body translation and rotation into the real leg coordinates using a rotation matrix.
When I find time for it I’ll help you out for simplifying the formula.
Are the coxa’s tilted and rotated exactly by 45 deg?
That would sure help since both sine and cosine of 45 deg gives the same result (about 0,7071).
I have been thinking about the 45 degree coxa servos and well…you can adapt the IK code easy enough to compensate for the 45 degree angle but even if you get the feet in the right position you’ll be introducing a side torque on the legs and chances are it’ll rip them off…or at least stress out your femur and tibia servos. Unfortunately, you need another degree of freedom to do it right. That means four more servos, more weight, and probably longer legs.
I’ll keep thinking about it. Why not “simulate” that 45 degree angle with some vacuum form parts and jsut mount the servos at 0 degree. poof problem solved.
hi yeah this makes perfect sense!
the coxa servo shafts are tilted down away from the body by about 43° as oppose to say 90° stright up… i can take a photo and add a diagram to the picture to show this??
once again the design favor’s function over aesthetic!
That looks fantastic! Can’t wait to see it walk.
