I have a mechanical rig similar to a human arm but the 'elbow' needs to hyper-extend past what the HIK preferred angle will allow. If I flip the angle 180 degrees, it moves in the correct direction however the elbow pivot is of course no longer in the right spot.
Any suggestions for how I can get this to work?
Thanks,
JD
Hi Lee, thanks for your response. Sorry my description isn't clear.
The vertical bar is connected to Box001. When it moves down, it pushes Box001 which will pull Box002 with it because Box001 is pinned in place at its center pivot. The two boxes 'retract' from the hyper-extended position about their linked pivot. The hyper-extension physically locks the mechanical rig in place.
If I detach Box001, Box002, change the HIK swivel angle to 180 degrees, it moves in the right direction but cannot hyper-extend which I need it to do. When I re-link the boxes to the bones, they of course detach at their conjoined pivot as shown below.
Let me know if you need some further details.
Thank you for your input, much appreciated!
JD
JD,
As I understand the mechanism, Box1 pivots about the fixed pivot point A. Box1 is connected to Box2 with a pin joint at B. What is not clear is what should control the angle Box2 make with Box1. For example, is point C of Box2 fixed to travel only horizontally? If not, what keeps C from moving down (or up)? Once the kinematics are clear we can determine whether it is appropriate to use an IK solver or perhaps an expression controller.
Lee
Hi Lee,
Point C will travel inwards to the left and up on an arc when the vertical bar moves down. B moves up and to the left on an arc while A remains pinned at all times.
This assembly pulls/pushes a flap open and closed 90 degrees. The flap is attached to C via a separately pinned cam of sorts at D.
Thanks!
JD
Is the action something like the following sequence? If so, where is the center of the arc and the radius for the path of point C?
Hi Lee,
The radius in my actual assembly is 1.623" centered at "D". May need to extrapolate for this example though :S. Point C in stage 1 below (pos1) will be hyper-extended a few degrees past where you currently have it. Pos2 is collinear to D.
Unfortunately I cannot post my actual assembly in this public forum.
Thanks!
JD
@jasond240 Actually I gave this problem a lot of thought and made some progress with a solution.
The essence of the challenge is to determine the angle of box2 (cyan box) as box1 rotates. The end of box2 should follow a circular path. The problem is that there is an inflection point where the continued rotation of box1 in a CW direction yields a solution where box2 can go in a CW or CCW direction.
To help me explore how to compute the angle of box2 I created a point A (PtA) that should be located at one of the two intersections of two circles. One of the circles is located at the pivot of box2 (yellow circle) and the and the other is centered at your point D (pink circle). I used a numerical approach (trial and error) to determine the appropriate intersection. In the code below, Po is an initial guess for the intersection point. It lies on the yellow circle. Pn on the pink circle at the intersection of a line from Po to the center of the pink circle. It is closer to one of the two true intersections of the two circles than Po. Pm is on the yellow circle and is at an intersection of a line from Pm to the center of the yellow circle. The calculation of Pm and Pn is in a while loop which continues recalculating Pm and Pn until the distance between Pm and Pn is less than 0.01 (an arbitrary error value). I placed a limit of 10 iterations on the loop to protect against the case when the two circles do not intersect. The process converges very quickly to a solution. The initial guess for Po determines which of the two circle intersection points will be calculated. My code needs some work to better choose the initial guess.
Slowly rotate box1 or box2 in the attached file (I have not tied the rotation of box2 to PtA) and you will see PtA move with the intersection point of the two circles. WARNING! The code will crash Max if the two circles do not intersect. My plan was to use PtA to determine the angle of rotation for box2. But at what point should the arm hyper-extend? My code jumps PtA to the other solution when box2 pivot is on the line from the pivot of box 1 to the center of the pink circle. This needs to be addressed.
Play around with my file and you may gain some more insight into the problem and a potential solution.
Script for PtA.
alpha = rotation of box1, beta = rotation of box2, P1 = pivot of box1. 8 = radius of yellow circle, cirRad = radius of pink circle.
beta = alpha + betaLoc P1= box1+[dum1.x*cos(alpha),dum1.x*sin(alpha),0] Po = [8.0*cos(beta),8*sin(beta),0] + P1 cont = 1 Pn = cirPos + cirRad/length(Po-cirPos)*(Po - cirPos) while cont > 0 do ( Po = Pn Pm = P1 + 8/length(Pn-P1)*(Pn-P1) Pn = cirPos + cirRad/length(Pm-cirPos)*(Pm - cirPos) if cont > 10 then cont = 0 if length(Pm-Pn) < 0.01 then cont = 0 ) -- end while Pm
Script for PtA.
Hi Lee,
That's some serious math going on there! I'll try to disect it. One thing I noticed is that box2's pivot should be in the middle, near your "box2" text rather than its end point. I'm afraid this may unfortunately negate the other circle and intersection point(s).
Below is a screen grab of my actual assembly, main pivot points/paths marked with circles. The top image (A) is correct (flap retracted), image B is deployed but the arms are not properly rotated as it still uses the original, non-functional bones. The red line in B is approximately where it should be. The pink & translucent parts move up and down, pushing/pulling the purple arm (pin travels in slot) to make it rotate.
Points on the blue and pink circles in the image travel their circumference in an inverse motion. Perhaps there is a way to constrain their pivots to a path and push/pull their joint (C)?
Bit of a head scratcher :S
JD
The problem as I now understand it is much simpler than I thought? I think a configuration like the following should work.
The cyan box (your purple link) has its pivot as noted in the top image. Three bones are created as shown. The root bone is linked to the box near its end and has an HI IK solver to the end bone. Rotating the box yields the motion you want! The flap (red, green and gray items in your drawing) are linked to the second bone in the chain. Your blue link is linked to the root bone. Keep the rotation of the link within reasonable limits and the mechanism should work fine.
