I’m looking at controlling an arm that closely resembles a humans. It will most likely be a hacked up mannequin arm and i’m looking to achieve roughly 120 degrees of motion (90 is acceptable but i will not go lower).
My problem with this is that the arm will move slowly and need to hold it’s position for up to twenty seconds. The torque i will require will be around 425 oz-in (3Nm). I require position feedback incase someone bumps my arm thus i’m looking at servos, i will also be controlling up to six servo’s at the same so the ssc-32 seemed ideal. The servo i was toying with was the hitec 5980 SG however it costs a bit more than i was looking to spend. Perhaps i will be able to make my own 2:1 gear system and just scrape in at 90 degrees of movement and use a cheaper motor.
Do you guys think that the servo’s will be up for holding this load still for up to 20 seconds at a time? If not then do you think stepper motors will be my best bet and do you know of any reputable stepper motor suppliers (controllers aswell if possible).
Alternatively I think that if I could find a good worm gear I would be able to use a full rotating servo (just count the rotations and have a neat program running it). Perhaps some of you robot veterans know of a good supplier of them?
I live in western Australia for the additional information.
I really hate to burst any bubbles, but I think this might be harder to achieve than you are expecting. First of all, a mannequin arm will most likely be way too heavy. If you calculated that torque with the weight of the mannequin arm in mind, it seems a little small (maybe a holding torque, but I assume you need to raise the arm?)
If you want to use a servo, you may want to check out ServoCity’s servo geared pan/tilts, even though by the sound of it, they are over your price range.
I think you are heading in the wrong direction with the stepper motor idea, I don’t think they have a lot of holding or driving torque for the price (and the price of a driver). A worm gear setup may be a good way to go, you’ll be able to get huge torque if you want it, but again, may not be cheap and a lot of work to rig it all up. Also, just for info, contiuous rotating servos have no feedback, so you would have to put an encoder or a pot or something on the joint.
I think your biggest challenge will be getting something slim enough to look human with enough power. Will be very interested to see what you come up with.
You may also find this interesting: lynxmotion.net/phpbb/viewtopic.php?t=2405
There is a link to a video on page 7 with Aiko moving her arms, you may want to ask animetown9 what he used for the arms.
On another note, I also live in Australia, and I’ve found that for stuff in this hobby, it’s usually more cost effective to buy international and have it sent here. I would prefer to buy local, but it’s so hard to find what I want, and for a reasonable price.
My intention is to make a frame body that i can hide via shirt/jacket and have arms poking out. I do agree, my torque values were made to be as small as possible whilst still being realistic but i am quite certain that 3Nm is achievable, 5Nm would be ideal.
I also knew about the requirement of an encoder however I’m not 100% on the prices of these, i didn’t even consider using a pot but will look into those also.
Thanks a tonne for the reply, I will send animetown9 a pm and try to hunt down some worm gears. I already am in possession of a ssc-32 but i think that if i chase the worm gear option and go for the continuous rotating servo + encoder i may be looking for another control card! (i’m going to be controlling 6 motors at once)
Question will be ‘what is cheaper’:
worm gear + lesser servos + encoder + control card
or
high end servo’s (roughly $800-$1,000)
or
a nice suggestion from animetown however it seems that price was not a large barrier for his project, it’s very impressive.
That is indeed a nice looking motor however the price of over $250 is much more than I’m looking to spend. The SSC-32 states that it requires Futaba or Hitec servo’s in order to operate also, have you guys had much luck running other brands?
First off, 425oz/in is more then even a 5990TG can produce continuously. Second, I believe the 5980SG doesn’t provide feedback.
Hobby servos are the wrong type of solution for this problem. You really need to look at worm gears and designing your own joints. Worm gears will also be fairly expensive, and you can expect to spend between $50 and $150 an axis just for the gearsets alone. Then you will have to add your own shaft encoder (pot, hall effect, optical), and your own motor and driver.
$250 per axis for something that strong is very reasonable. (and honestly probably on the low end of a DIY solution)
Personally, I would look at:
Worm Gears
Standard Brushed DC motors
A encoder that outputs 0 to 5V analog (either a pot or look at US Digitals hall effect)
Polulo (sp) Robotics intelligent speed controls that can accept a 0-5V shaft position and PWM servo signal and control a motor from that.
Much appreciated Sienna. If i cover all my options and the joints are still going to out do my budget ($100 per joint) (and my alternate funding still won’t cover it) i will have a compelling argument for a change in the original project.
I sent an email to Hitec just before my initial post here regarding torque vs speed information. I had a strong feeling that the motor wouldn’t be able to offer the torque i needed (especially in the holding position) but wanted to double check. (no reply as of yet)
I’ll keep on hunting for the time being and I will let you know what my final solution is. If i still end up running the SCC-32 control then chances are you will be able to see the evolution of the project. Feel free to throw any further information up because it is still useful to me.
It’s the age-old battle between torque and rpm for a gear-reduction system (servos). The only way to offset this is to use a stronger, bigger, and usually more expensive motor inside the servo.
If your application demands torque and speed, usually you’ll have to pony up some dough to get the job done.
Of course, if you don’t need the degree of precision and just want to create some animatronics, you can use a motor and pulley system.
When it comes to robotics, there’s no way to do it on the cheap when you want performance in your application. A lot of biped humanoid people out there that does this for competition have spent enough money on their hobby to buy couple of cars
More DOFs, more servos, more requirement of torque, more money… Sigh…
In my application speed doesn’t bother me to much. I need to achieve about 90 degrees in one second (15rpm) so the mega gear ratio is just fine.
I will be focusing largely on the control of these motors so accuracy is an issue, that was one of the largest faults with the stepper motor because i was feeling lazy and didn’t want to implement an encoder on it! (people might bump the arms leaving them in strange positions if i don’t have feedback). Torque and accuracy but without speed miiiiight be possible with this worm gear idea. I need to go learn some theory to work out how strong the gears need to be, it could save allot right there and let me get a more suitable motor .
A reasonably hard project with a restricted budget with 6 months until completion. This is going to be a good learning curve for me without a doubt!
hey heres an idea…why not use a light weight frame made of small strips of plastic tubing …like used in the r/c planes…these are very light and flexable…also can be used to add distance to a remote arm or hand so the motors or servos can be off the thing your tring to move .
as for motors you can use drill motors and a spring as a worm gear to control them use a r/c to motor contrioller like a sabertooth.
ck out this vid of a project I’m helping on…these are legs for a 4 foot hexpod.
You can show me the plastic sure, i don’t think flexible is what i’m looking for because the arms will remain rigid however light weight is exactly what i need.
The spring idea is quite impressive but i imagine that under higher loading it gets pretty useless as the spring coils begin to compress? What ratio did you manage to achieve from that also if you don’t mind sharing. I’ve found a nice little worm gear set up from a local hobby store however the gears are plastic and i have some serious doubts (still havn’t found my theory yet) that they will be able to hold the load without threading.
I have two of those… and for being plastic, they are actually pretty tough. Now, I have no idea how strong they are torque wise, as I haven’t tried to destroy them (too hard - the set screw came loose before I was able to damage the plastic)
Unless it is a deep dark secret, you probably need to give more details on what you are expecting from the arm. You might consider using linear actuators as they are strong and could some what be set up like a large muscle is attached to bones. Does the arm pick up objects, and if so, what is the weight of the objects? How many joints are you expecting to have in the arm? What level of accuracy do you expect?
I dont think that a dummys arm is the ticket…
What I was refering to was to make a “shell” so that you could put a shirt over it…looks like a solid but is really light…go to the local hardware store ,they usually have some kind of lenghts of aluminum tubing and strips to get the rigidity you need…
You guys have been very helpful so I’ll elaborate on my project. It’s not really that secret. Essentially i will be making a life sized demonstration mannequin. The original plan was to have the mannequin track and point at people walking past (through the use of a camera) however due to time restrictions i have changed the project already to a simple computer program control that allows the user to select a location for the mannequin to point at.
The shoulders will have two degrees of freedom, the elbow will be locked, the neck and waist will both rotate horizontally.
It will NOT carry additional weight however chances are someone will grab it at one stage or another and it will get a bit more force than planned.
It should point to the desired location and not be out by more than ± 4 degrees.
Speed isn’t a huge issue but it can’t put people to sleep whilst it moves. I plan to have the arm sit in the desired location for around 10-20 seconds before returning to a rest position.
I do not have a fixed weight so my calculations are based on estimated loads.
Using those light rods or similar along with a human style hand + a shirt to cover the rod is looking to be a very good idea.
I still like the worm gear idea because it will allow the limbs to stay in position without motor wear/ power waste however the combination of a gear box and reasonable servo can theoretically produce 12Nm of torque whilst still moving fast enough to be of use and costing under $100. It just requires a bit of a conversion of the servo which i have not done before. *thank you very much animetown9 for the suggestion.
PS. Sienna those gear boxes will take a couple weeks to come in because the manufactures are busy over on the other side of Australia some sort of expo (the stanbridges guys had a chat with me). I did manage to find the information on the gear box and i would have to replace the motor and push the gears 10x harder than the old motor would have so it’s not worth the risk.
You might be able to operate the arm using 1/4 scale or less servos, and make the arm itself out of a styrofoam base material. Use something like roller skate wheel bearings at the rotation points for strength to take loads off of the servos. Actually sounds like it could be fairly simple to make if the arm only moves at the shoulder.
The outer arm motor should be relatively simple but the inner shoulder that has to support both the outer arm motor and arm itself is what has been giving me stress due to the load. I will look into 1/4 scale servos and the styrofoam could actually be better than the rods. I can’t remember where but i do remember seeing a styrofoam dummy before, that would reduce my load hugely whilst still looking realistic (one some paint is applied).
As this is the first larger scale project i have worked on it is still a large learning curve for me which is why i’m finding out as much as i can about actual prices (never been an issue before) and your help has been great.
I suggest using the servo brackets that include the ball bearings on the servos. A bearing supporting the bracket on the servo side opposite the servo horn would make for a fairly strong movable mounting point. For the arm, use a servo mounted on its side such that the arm is attached to the bracket and the servo moves it up and down. To rotate that servo left/right, mount the arm servo on a servo bracket of a verticle servo. The body of the verticle servo would be mounted to the shoulder section of the bot.
I imagine that my brackets will be custom made, I just need to size up the motors before i put the final dimensions on my frame (already designed mostly) . If i can find brackets that will do exactly what i need then i’ll think about getting them but cost cutting is a big thing for me as you’ve seen!
.