Results 1 to 8 of 8

Thread: How does one set the torque of the servos for a moving robot

  1. Question How does one set the torque of the servos for a moving robot

    Hello ,


    I'm trying to make a biped using the Dynamixel Ax-12A servos for the legs. In the User manual for these servos it is clearly stated that:


    Stable motions are possible with robots designed for loads with 1 /5th or less of the stall torque


    Does the above statement apply to just 1 of the AX - 12A Servos or should the total mass of the robot be 1/5th of the total stall torque of the 6 servo motors (3 for each leg) i.e. all of them added together, during the motion of the robot.

    Any help will be much appreciated

  2. #2
    Join Date
    Sep 2008
    Location
    Toronto, Ontario
    Posts
    171
    Images
    43
    Rep Power
    40

    Re: How does one set the torque of the servos for a moving robot

    That's the mass of the robot to 1/5th the stall torque of one servo.


  3. Re: How does one set the torque of the servos for a moving robot

    Thank you so much for replying Kamondelious

    Quote Originally Posted by kamondelious View Post
    That's the mass of the robot to 1/5th the stall torque of one servo.

    Does this mean that the if I'm using an AX-12A servo (which has a stall torque of 15kg-cm), then the weight of my entire robot should be less than 3kg?

    To cut everything short, I want to use the Ax-12A servos, 3 of them in each leg and 2 in the hips, so that it's strong enough to lift a mass of 3.5 - 4 kg.Will the these servos be suitable for this application.


    Thanking in anticipation.

  4. #4
    Join Date
    Sep 2008
    Location
    Toronto, Ontario
    Posts
    171
    Images
    43
    Rep Power
    40

    Re: How does one set the torque of the servos for a moving robot

    Well, based on the quote from the user manual you provided :
    Quote Originally Posted by Aims Bot View Post
    Stable motions are possible with robots designed for loads with 1 /5th or less of the stall torque
    and my own experience, yes it would be possible, but it's very much dependant upon a few things (not listed in any particular order).
    1) How stable does it need to be?
    2) How long does it need to be able to walk for?
    3) How fast are you expecting it to move?
    4) Is the leg design well suited to lifting?

    If all you're doing is lifting, you should be ok, provided you are not trying to hold positions this maintain strain on the servos. Also that you're not looking for very long run times without adequate cool down times.

    You may also want to have a look at PLM (Parallel Leg Mechanism) style designs.

    Can you share a little more about what your project is?

    Cheers,


  5. Re: How does one set the torque of the servos for a moving robot

    Hi,
    Thank you soo much for the prompt reply.
    Yes, so my project is about building a biped that can lift weights on the shoulders and the back( just like a robot is carrying a backpack)
    But these weights won't be attached to the biped at a time. We can do that one after the other. Wherever we decide to put the weights, they will be attached firmly. These weights can be between 2 - 2.5kg. And the weight of the biped itself will be 1.5 kg (including the servos, servo brackets, and other components like microprocessors). And all the biped is supposed to do, is lift this weight and walk around.
    Now, I felt that by using the Ax-12A servos I'd be leaving very little margin for myself, so i have decided to use the Ax-18A dynamixel servos. These servos according to the user manual has 18kg-cm. And if I act along the lines of the previous statement from the dynamixel servo manual:

    Stable motions are possible with robots designed for loads with 1 /5th or less of the stall torque

    then it means that the total weight of the robot now can be the 1/5th of 18kg-cm i.e 3.6kg and the total weight of my entire unit along with the weight at one of the places will be 3.8kg. 0.2kg greater though.
    Will the biped be able to walk stably?

    The speed of the robot should be the same as any average biped's speed or even a tiny bit slower and it's run time shouldn't be more than 2 to 3 minutes.

    Will the Ax-18A be suitable for this application?

    Now, I'll be using arduino as my main microprocessor and I've seen a few libraries for the Ax-12A servos in arduino. So, another of my concerns regarding Ax-18A sevos is that, are they compatible with arduino? If, so could you please elaborate on how?
    Is there any arduino interface that can be used for the Ax-18A servos? Can I just use Ax-12A arduino libraries for making the Ax-18A servos work as well.

    I really really appreciate your help.
    Thanks a million.

  6. #6
    Join Date
    Sep 2010
    Location
    ಠ_ಠ
    Posts
    2,313
    Images
    27
    Rep Power
    278

    Re: How does one set the torque of the servos for a moving robot

    The maximum continuous torque delivered by the servo without a significant risk of rapid overheating is ~1/5 of the stall torque. You will have to determine the loading (force * distance from center of horn to the perpendicular of the line of application of the force) of the servo to know whether it is exceeding the usable torque. To do that, you will have to measure the worst case distances in your gait, which will approximately be the horizontal distance between two adjacent servos in the chain. That is enough for a rough estimation, since better values will require a free-body diagram and calculation of the centroids of the various masses in the torso in each pose of the gait. e.g. the Bioloid Premium Type-A would have worst case torque on the knees, hip-pitch, and ankle-pitch servos when attempting to hold the torso vertical with the knees bent so that the hip is held just above the ankle (a gap of a couple millimeters between the two pairs of servos). The lengths of the lever arms under that condition would be roughly the horizontal distance between the point of rotation of each servo (knee to hip, and knee to ankle), and the force would be the weight of the robot and payload. The lowest loading on those three servos would be when the knee is vertically aligned with the ankle, hip, and torso ('locked' straight like a human knee), which is a large part of why most organic bipeds don't walk with their knees constantly bent to small angles between the upper and lower leg like the DARwIn-OP gait (energy efficiency of the locked knee, swinging leg gait is such that completely passive bipeds can walk down a ramp of only a few degrees).

    The AX-18A have a coreless motor that is faster and higher torque than the AX-12A, but the smaller thermal mass of the rotor can cause them to overheat a bit more quickly than the AX-12A. The AX-12A and AX-18A are identical in being controlled by any dynamixel library for arduino. The only difference the software might notice is the firmware version and model number being different (most libraries completely ignore both of these).
    Please pardon the pedantry... and the profanity... and the convoluted speech pattern...
    "You have failed me, Brain!"
    [git][mech][hack]
    gives free advice only on public threads

  7. Re: How does one set the torque of the servos for a moving robot

    Thank you so much for such a detailed reply. It was really helpful.


    I had a few doubts though:

    To do that, you will have to measure the worst case distances in your gait, which will approximately be the


    horizontal distance between two adjacent servos in the chain

    Do you mean the maximum horizontal distance between the two adjacent servos on the legs of the biped? or the distance
    between the two servos on the same leg? Could you please elaborate a little on this method?


    There is another method that I'm using to estimate the endurance to torque of a servo is as follows, please don't hesitate to
    let me know if you could correct or add to it,

    So, if the torque of the servo is 18 kg-cm (AX-18A, dynamixel) and the hip servo (I'll be using two servos in the hip) is suppose at 20cm of height; does that mean that the weight this hip can withstand will be equal to:
    18/20 = 0.9kg
    and as the height decreases to the ankle servo, the weight these servos can withstand increases, because the height at which these servos will be present decreases. Hence, finally the total mass one of the legs can carry will be the sum of the weights the hip, knee and the ankle servos can carry altogether, right?


    Also could you please tell me a method of finding the the centeroid of the robot, when it has the masses present at it's
    back or on it's shoulders? Or the centeroid of the torso?


    Could you please suggest me the led, arm and hip assembly for my robot, using the AX-18A servos along with the dynamixel
    servos?


    As from you're post I've perceived that the locked knee mechanism leg bears the least amount of torque , so I'm going with that.

    For my robot to stand up again once it falls down, along with lifting 2- 2.5 kg at it's back or on it's shoulders, will the Ax-18A servos be powerful enough to perform this function?


    Your help will be very much appreciated.
    Thanking in anticipation.

  8. #8
    Join Date
    Sep 2010
    Location
    ಠ_ಠ
    Posts
    2,313
    Images
    27
    Rep Power
    278

    Re: How does one set the torque of the servos for a moving robot

    http://www.societyofrobots.com/mechanics_statics.shtml <--Go there. Read it.

    So, if the torque of the servo is 18 kg-cm (AX-18A, dynamixel) and the hip servo (I'll be using two servos in the hip) is suppose at 20cm of height; does that mean that the weight this hip can withstand will be equal to:
    18/20 = 0.9kg
    and as the height decreases to the ankle servo, the weight these servos can withstand increases, because the height at which these servos will be present decreases.
    horizontal =/= vertical. Think of yourself keeping your torso vertical while: 1) standing up with your knees locked straight, 2) squatting with your knees bent at 90 degrees. With knees locked, the vertical force due to gravity is transferred vertically through the leg bones, so that you can mostly relax the muscles in your legs and abdomen yet stay standing. When squatting, the horizontal distance between the knee and hip/ankle (and open space directly beneath the centroid) requires an additional torque in the knees, ankles, and hips to maintain the pose (to prevent the hinges/joints from swinging freely). Let the muscles relax at all and you will fall down.

    Hence, finally the total mass one of the legs can carry will be the sum of the weights the hip, knee and the ankle servos can carry altogether, right?
    NO. 'Only as strong as the weakest link'. The payload is not the sum of the torque limits, but is limited to the maximum of the weakest joint/servo in the kinematic chain in a given pose (leg supporting some portion of the torso mass). If the robot has two legs, the weight of the torso can be divided between them when both feet are on the ground, but a single leg still must support all the weight when it takes a step.

    For my robot to stand up again once it falls down, along with lifting 2- 2.5 kg at it's back or on it's shoulders, will the Ax-18A servos be powerful enough to perform this function?
    Again. It depends largely on how you design the robot, but a Bioloid Premium Type-A Humanoid with AX-18A servos would probably not be able to do that without overloading and/or overheating.
    Please pardon the pedantry... and the profanity... and the convoluted speech pattern...
    "You have failed me, Brain!"
    [git][mech][hack]
    gives free advice only on public threads

Thread Information

Users Browsing this Thread

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

Similar Threads

  1. Replies: 4
    Last Post: 12-09-2014, 09:01 AM
  2. Question(s) Controlling lots of Imperial servos on a Dark side robot?
    By Slugman in forum Robot Computers
    Replies: 8
    Last Post: 01-13-2012, 06:25 PM
  3. How to increase torque by paralleling multiple servos
    By bonmot in forum Humanoids, Walkers & Crawlers
    Replies: 6
    Last Post: 03-26-2011, 06:48 PM
  4. C# code for moving servos!
    By bayyari in forum Software and Programming
    Replies: 3
    Last Post: 09-11-2010, 08:33 AM
  5. Wanted Hi torque Servos
    By Meccano Jay in forum Garage Sale
    Replies: 3
    Last Post: 02-25-2010, 05:20 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
  •