1
$\begingroup$

I'm interested in exoskeletons and wearable rehabilitation robotics. I wonder how we can estimate/predict the intention of human body/part motion. I want to prevent the exoskeleton from interfering with human movements. Intention reading is the process of predicting how the movement will take place and how it will happen at the beginning of any movement.

There is an exoskeleton example (https://www.youtube.com/watch?v=BdoblvmTixA) which detects muscle activation with EMG and generates artificial muscle attraction. But this is only a open and close action. And it will begins after the movement. Also EMG system has a lot of disadvantage like sliding probes, affecting from other/crossing muscles. I want to estimate every motion like turn,twist, amount of contraction. I'm open to suggestions or issues (The troubles you have experienced.)

This matlab webinar which is about "Signal Processing and Machine Learning Techniques for Sensor Data Analytics" shows how to classify different actions. But this example predicts kind of motion after the motion completed. I need to know motion information at very first.

I want know how can I estimate different motions at the beginning of limb action. Which system (EMG,EEG,IMU,etc.) and processing technique will be better or which combination should I use.

$\endgroup$
  • 3
    $\begingroup$ This is a rather broad question which also solicits opinions, so could be closed for either reason. To make it a practical, answerable questions based on actual problems that you face it would be a good idea to include details of what you want to achieve, how you tried to do these calculations, what you saw & what you expected to see. Remember the Robotics question checklist is there to help you edit your question to make it clearer. $\endgroup$ – Mark Booth Dec 12 '16 at 16:29
  • $\begingroup$ @MarkBooth question edited. $\endgroup$ – acs Dec 12 '16 at 22:23
  • $\begingroup$ Thanks @acs hopefully the extra detail in the question will elicit more comprehensive answers. $\endgroup$ – Mark Booth Dec 13 '16 at 9:53
  • $\begingroup$ Are you interested in measuring motion or measuring intent? Those are two completely different things. Measuring motion needs to happen around the joints, with any number of motion capture techniques. Intent would be more difficult (as you outline) and, like any other signal transmission, requires capturing the signal before transmission (brain wave), after reception (muscle potential), or somewhere between the two (nerve/spinal interception). But again, what do you actually care about? Intent or action? $\endgroup$ – Chuck Dec 13 '16 at 16:45
  • $\begingroup$ @Chuck Your detection is correct. Intent reading will be a more accurate definition. $\endgroup$ – acs Dec 13 '16 at 21:35
3
$\begingroup$

You have many options: encoders at exo joints to provide absolute angles, IMU on limbs for estimating motions, force sensing shoe soles for force distribution, force sensors at joints, crutches usually used by pilots wearing rehab exo can also have those sensors, e.g. at the bottom of the crutch, at hand handles, etc.

$\endgroup$
1
$\begingroup$

MYO gesture control looks like a good system at least estimation on arm and hand motions. It includes multiple EMG and IMU sensors.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.