Hey All:
Major robo-noob, here. I am a PhD student studying human kinetics. Currently, I am studying the control of human elbow movement. We strap peoples’ forearms into these leverlike things with a fulcrum that is on the same axis as the elbow joint and measure the incline using MEMS accelerometer chips…imagine two side-by-side half see-saws at the edge of a table with the elbow at the fulcrum and the hand at the far end. I use MATLAB programming environment and an A/D interface to record the angles from the MEMS, and to flash little lights at the subjects. Right now, I am studying what happens when people actively move their own arms, but I want to study what happens when people have their arms moved passively. I will be moving the arm through 90 degrees of movement, at very specific rates of speed, to very specific positions. So I have a few questions:
I want to use either a linear actuator or a stepper motor to move the subjects’ arms in a controlled manner. I noticed that most steppers say +/- 5% precision. Is that 5% of the 0.9 degree half step mode or 5% of the total 360 degrees? Also, do you mean 5% precision, or do you mean 5% accuracy? In my books, precision is about repeatability, and accuracy is about actual distance from a desired target.
A forearm weighs around 6 pounds (give or take). I would guess that the 125 oz/in torque motors robotshop.ca/rbsoy14-soyo-unipolar-stepper-motor.html would do the trick, but I am not sure. If I used a linear actuator, I would guess one of the Firgellis at 15 lbs and 8 inches would do the trick robotshop.ca/firgelli-automations-15-12-8-mini-actuator.html. There is an issue of safety: if I use a very powerful device and it doesn’t stop where it should, I want the device to fail, not my subject’s arm.
I want to connect the stepper or actuator to my A/D, and then control it and monitor through my MATLAB software. I have no idea if any of this is possible. Do stepper motors or actuators provide feedback of their location, or do they need to be sensed externally? Can one control rate of movement? How accurate might they be?
Any help at all would be really appreciated. I know this is kind of outside of the ordinary, but hey, you guys have the motors and actuators.
cheers,
Blake
Mr. Barnes, I am doing a PhD in Kinesiology, with a graduate diploma in neuroscience. I am studying how attention interacts with how your brain figures out where your body parts are when you do more than one thing at once. I like to say I am studying “how do we walk and chew gum at the same time”.
In reality, the problem I am studying is very much like the problem we are discussing: You have a central plan, such as a movement created by your brain or some microprocessors and code. You have an effector, either an arm or an actuator. You have a goal position or movement. You have external factors such as load or obstacles. And you have a feedback system to tell you if the movement worked and a means to fix it if it doesn’t work.
The robotic arm calculator is telling me about 22 N-m. Seems a bit high. I didn’t enter anything for the mass of the actuator since it is stationary and supported.
Jonathan, from what I can tell, servos a) look way more expensive than steppers, and b) don’t look as powerful. But from what you are saying, they have internal feedback (forgive me if I am saying this wrong). I don’t need wireless. I just need to put the subjects arm into several positions, sometimes slowly, sometimes more quickly. I have to get the arm to the same places everytime. A PC will be used to handle the logic, and separate tilt sensors can be used for feedback. Neither of you mention linear actuators.
Since I am a student, I want to keep this cheap. Since I am not an engineer, I want to keep this simple. The fewer subsystems I have to learn about, the better.
Thanks for your responses.
Blake
You definately need to start by determining how much torque would be required on each arm axis, we have a tool to help with that. People usually want a geared motor with feedback on robot arms, many stepper motors will be insufficient to direct-drive a robotic arm:
Robot Arm Torque Calculator
Once you know how much torque is required, you can start looking at suitable actuators for the job and how to control them to meet your goals.
That is usually what people do with Steppers. The problem with steppers is that since you know the position by the number of steps you tell it to make, if an outside force changes its position, you are now lost. Hitting a reset point will help but ideally optical encoders, a simple potentiometer or even magnetic encoders should be used for position feedback. Of course it is much simpler to source an actuator that provide this feedback loop internally which is why servomotors are so popular in robotics.
You can go linear actuator, or any other actuator technology. The key is determining how much torque/force you need, the duty cycle, etc. and then determine if the actuator will be sufficient. You will assuredly need to experiment.
Keep in mind most standard linear actuators with position feedback actually incorporate a DC motor, pot and mechanical edge of some sort.
If you look at the motors used on industrial robot arms they are prohibitively expensive for hobby clients but they are required for lifting heavy loads repetitively and reliably.
If you want cheap and easy to use, standard hobby servo motors are a good choice. Tell them to go to a position and they just go there, repeatedly and reliably within their technical limitations. They are probably not the best choice for your project, which involves manipulating a person’s arm.
You wouldn’t want to break the subjects arms I’m sure. Any time a human is involved, you have to choose carefully since any failure could cause harm or even death.
What are you doing your Ph.D in?
As for a fail-safe mechanism, your best bet would be to use hard stops.
