You're right, Rsdoran, I should really have explained the senario a little better. First, let me handle a few issues.
First of all, the mayority of the guys contributing to this forum ARE professionals in their field and frankly deserve some respect. Therefore, I assume most replies are based on misunderstandings or lack of information (on your behalf).
I agree, although the post to which you are refering was soley directed at Rsdoran's previous post. If you can not agree that it was slightly Malevolent then I don't know what to say.
...With 'servo' I think you are referring to the compact steering modules used on modelling cars, etc. rather than servo motors used on handling equipment or industrial machinery. This would explain why your servos burn, since they are intended to try and reach their setpoint, no matter what obstacle comes along: There's absolutely no torque control whatsoever!
Initially that was all we understood of servos. We were given a servo similar to one out of an RC car which uses a closed-loop system to achieve a specified input (in degrees.) Now, however, we have a much more broad scope of what servos are and can be. We are currently looking into making our own using a helical gear box, dc motor, and a series of boards for gain and controls.
And thus the application comes out. That should steer the responses some. Is this arm limited in rotation or can it spin indefinitely? It sounds like this will be interacting with people. Do you have maximum force/torque and speed numbers nailed down?
What was the failure mode on the servo? Did you yield the gearbox or did you fry a motor or drive? Or did the arm break because the back drive forces were too high? What gearbox type and reduction were you using? Why are pneumatics not safe enough? Does the axis move too fast or are the forces not controllable enough? Is this assessment based on a design iteration or is it based on a design review? What have you tried with pneumatics?
The project is simple in nature, we are to construct an arm for the robot called RUBI, currently used at UCSD's Early Childhood Education Center and developed at UCSD's Machine Perception Laboratory. An article on RUBI can be found
here.
The details concerning RUBI are not a big factor, but the arm must be very specific.
The first attempt at an arm used 7 of the before mentioned "RC Servors" although they were a higher end device. We do not no for certain what part burnt out, whether it was the gearing or an electrical failure, but we do know why they failed. After only 2 days of interaction with the children, all 7 of the servos failed simply because they were not meant to be back driven.
The arm we have been assigned to create will only have 3 degrees of freedom: A 1 DoF pivoting shoulder, a rotating forearm (180 degrees of rotation), and an expanding/contracting hand. What we are focusing on is sole the shoulder and what type of actuator to use for this component. The shoulder only needs to rotate about 90 degrees, and it does not have to do so very quickly. There are no design constraints at this point in regards to velocity or torque, it is all up to us. The goals of the arm are as follows:
1. Lifelike compliance - We would like the arm movements to simulate a real person. The action we would like to simulate can be imagined by swinging your arms back and forth as if you were power walking, although not nearly as fast, simulating a sort of 'dance.'
2. Robustness - The arm needs to survive. The goal is 100 hours of use without breaking. As I mentioned, the servos on the last arm were destroyed in 2 days flat. This is where our professor began to use the term "back drivable." He would like us to use some sort of servo system that, when traveling one direction, can be pushed backwards without causing harm to the motor or gearbox. He was not specific at all in what back drivable meant, and that is where the issue began.
3. Safety - The arm has to be safe. The children it interacts with are between 18-24 months old. We have to assume the worst: Pinching, falling, grabbing, pulling. Anything that can happen will happen around children. So far we are only considering that kids will be often grabbing and yanking on the arm, or accidentally falling on it. When this happens, one option is to simply have the arm give (i.e. be back driven.)
As far as pneumatics go, so far we have developed a 1 DoF simulation of the shoulder powered by a
Vex Robotics pneumatic system. The main problem with this pneumatic is that the piston diameter is so small that the volume of air in the cylinder is easily compressed, and thus creates a large 'pop' when released. This action is dangerous to say the least. I believe that we can counter this by using a larger diameter cylinder, thus allowing less of the piston's stroke to be manually compressed, and limiting this pop. My professor on the other hand thinks pneumatics are not the way to go and suggest we research back drivable systems. Too bad no one knows what they are
I probably left alot out so if you have any more questions that would help you in helping us, feel free to ask. I will try to check this board alot more often now. Sorry for the attitude previously, it was just very frustrating when we were saying all we knew and still getting nothing. Thanks!
