Results 1 to 6 of 6

Thread: Can I use a steel ball to emulate a virtual floor with infinitesimal contact point?

  1. #1

    Can I use a steel ball to emulate a virtual floor with infinitesimal contact point?

    On the previous iteartion of Onyx, I used narrow-ish aluminum leg contact points for the "toes" that actually made contact with the ground. I think they were about 1/4", maybe 8 mm.

    I noticed that this actually changes the IK solution around a little bit -- my math didn't entirely compensate for this, and thus the solution was 3-4mm off, depending on which slant the leg made contact with. (I only had 3 DOF in the leg, so the angle was not under control.)

    Suppose I use an aluminum tube as my outer tibia/toe, and I glue/epoxy/braze a stainless steel ball onto that tube.
    An observation is then that the ball will always make a perpendicular contact to the floor -- from contact point, to center of ball, will always be the radius of the ball, and will always be perpendicular to the floor.

    So, it should be possible to just pretend the floor is one ball radius above the actual floor location, and then do the IK to position the center of the ball, rather than the actual floor contact position!
    I feel like I've been thinking in that direction before (I rounded the outer edges of the tibia, for example,) but I think this is the best description I've come up with, and with the simple ball-on-tube construction, it seems like this is all coming together very well.

    I hope I'm not missing something here?

  2. #2

    Re: Can I use a steel ball to emulate a virtual floor with infinitesimal contact point?

    A-ha! But when I actually move the leg, if I assume the ball center is stationary, then the ball contact with the floor needs to slip.
    Or I need to add a rolling mechanics compensation that knows what angle the tibia is touching the ground at.
    Or I go back to the idea of sharp-spike steel feet.

  3. #3
    Join Date
    Jun 2013
    Location
    Tucson, AZ
    Posts
    253
    Rep Power
    32

    Re: Can I use a steel ball to emulate a virtual floor with infinitesimal contact point?

    On my mech '99 running IK, I just use a rounded foot similar to your ball bearing idea. At slow walking speeds, I see almost no discernible effect, if there is any on the IK. At higher speeds, the dynamics of the actuators come into play and you don't get the perfect IK solution.

    Heck, when I first started on '99, I used an F4 bracket for the leg/foot. It put a lot of roll into the video, but the pitch/yaw were pretty solid.

    A couple of the other mechs running IK in our club use feet with ball joints. They just consider the center of the ball joint as the ground location.

  4. #4
    Join Date
    Jun 2013
    Location
    Tucson, AZ
    Posts
    253
    Rep Power
    32

    Re: Can I use a steel ball to emulate a virtual floor with infinitesimal contact point?

    Out of curiosity, how far along are you on your IK code? I wrote the IK code that our club uses and it was not a trivial task.

  5. Re: Can I use a steel ball to emulate a virtual floor with infinitesimal contact point?

    Using a ball as the foot is a common technique. However usually people us something that grips, not a steel ball.
    The assumption that you use to simplify your IK computation of a perfect contact point does not hold if the foot slides by lack of friction/striction.

    And compensating "well enough" for the ball rolling when the leg changes orientation after making contact is not too complicated after what you have done to get the IK solution: compute the inital configuration (angle of the leg from vertical) when the foot makes contact, then when you want to move, recompute that angle, and use the difference from the original configuration to compute how much the feet has rolled on the ground, and then project on the ground plane to know how your ik solution is affected.
    You don't necessarily need the exact solution that wouod require computing IK and that offset at the same time, you could use compute the offset from the previous IK solution, assuming you do compute it regularly and fast enough compared to the dynamics of your robot (i.e. if you only do 3 or 4-pose walk cycles, that won't work, but if you do continuous interpolation it would be ok).


    Also, on a more pragmatic approach, if your foot is 8mm in diameter and your robot in the same general size as your previous ones, we could argue that backlash and structural part torsion/flexion will already have a much greater impact than the ball of the foot rolling without slipping. So unless you plan to model that too or to us a bigger foot, correcting for that minute systematic error is probably not going to have any visible impact.
    Last edited by Xevel; 11-07-2016 at 04:21 AM.
    ---
    Personal blog: http://xevel.org
    USB2AX documentation: http://xevelabs.com

  6. #6

    Re: Can I use a steel ball to emulate a virtual floor with infinitesimal contact point?

    Out of curiosity, how far along are you on your IK code? I wrote the IK code that our club uses and it was not a trivial task.
    Yeah, it took a couple of days to get right! But I got that right like two years ago, and it's been working fine since then.

    we could argue that backlash and structural part torsion/flexion will already have a much greater impact
    I was in fact having that argument with myself while posting the initial post :-) I'm more interested in the actual math here.
    It's true that estimating the amount of roll isn't that hard when looked at in that way. The hard part is knowing which of the legs actually slipped. when you have multiple contacts that can't all be 100% fixed!

Thread Information

Users Browsing this Thread

There are currently 1 users browsing this thread. (0 members and 1 guests)

Similar Threads

  1. Real Steel movie
    By jdolecki in forum Mech Warfare
    Replies: 0
    Last Post: 09-27-2011, 07:32 AM
  2. Question(s) anyone have Lil-Brother LLC contact information?
    By societyofrobots in forum Robotics General Discussion
    Replies: 0
    Last Post: 05-22-2010, 03:09 AM
  3. Best entry point into Robogames
    By Resilient in forum Robotics General Discussion
    Replies: 8
    Last Post: 02-28-2009, 12:11 AM
  4. Virtual Testing
    By Firestorm65 in forum Mech Warfare
    Replies: 5
    Last Post: 02-21-2009, 11:45 PM
  5. Another guy we'll need to contact
    By ScuD in forum Mech Warfare
    Replies: 9
    Last Post: 10-17-2008, 06:44 PM

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •