Using InMoov for a Prosthetic Arm

[Brancante]

Hello All, 

I have been playing around with an EKG/EMG Shield to detect a muscle movement to control the InMoov Hand.

 The progress is good, but needs a lot of work still.

 This is the first video I have made: http://www.youtube.com/watch?v=ZUIqR2JodlY

 

I am using the SHIELD-EKG-EMG from olimex: https://www.olimex.com/Products/Duino/Shields/SHIELD-EKG-EMG/

But this is a single Channel EMG and I am having a hard time decoding advanced movements. I have ordered a different shield with four channels.

I will continue the development and will post the updates and Tutorials.

[gael langevin]

Fantastic Brancante!
Didn''t expect through your email the other day, how advanced you were. I' m sure the Fablab of Rennes France will be happy to see this. This is full of promises!

[Brancante]

Thanks Gael.

As promised, here is a new video and the tutorial.

http://youtu.be/38SnocWQFkM

 

I have implemented new features, now I can make multiples movements controlled by the muscle with combination of specific modes. The modes can be configured by a smartphone and wifi.

 

Here is a small tutorial:

 

What you need:

 

InMoov Arm

Arduino uno R3

EMS Shild - https://www.olimex.com/Products/Duino/Shields/SHIELD-EKG-EMG/ (

Wifi Shield (UDP compatible)- http://www.ebay.com/itm/UART-WiFi-Server-Client-Module-Kits-Arduino-Compatible-with-shield-for-Arduino-/130766449058?pt=LH_DefaultDomain_0&hash=item1e7249a9a2

Software TouchOSC (for smartphone and Workstation) - http://hexler.net/software/touchosc

 

The Wifi Module should be configured as UDP Server with fixed IP (ip and network from phone hotspot) and 57600 bps

 

This is the TouchOSC Layout:

https://dl.dropboxusercontent.com/u/14231908/EMG.touchosc

 

 

This is the code for Arduino:

 

  #include <Servo.h>

  #include <compat/deprecated.h>

#include <FlexiTimer2.h>

#define SAMPFREQ 256                      // ADC sampling rate 256

#define TIMER2VAL (1024/(SAMPFREQ))       // Set 256Hz sampling frequency                   

volatile unsigned char CurrentCh=0;         //Current channel being sampled.

volatile unsigned int ADC_Value = 0;    //ADC current value

volatile unsigned int ADC_Value1 = 0;              //ADC current value

 

Servo myServo1;

Servo myServo2;

Servo myServo3;

Servo myServo4;

Servo myServo5;

Servo myServo6;

 

  int pos = 0;    // variable to store the servo position

  int modo = 0;

  int oscMsg[12];    // buffer for incoming OSc packet 

  int inByte = 0;    // incomming serial byte 

  int inbyteIndex=0;   // incomming bytes counter  

  float msgVal=0;    // resulting message value 

  void setup() 

  { 

 

 

   Serial.begin(57600);   // debug port 

  

 

  

myServo1.attach(13);   // pulso

myServo2.attach(8);   //dedo mindinho

myServo3.attach(9);  //dedo2

myServo4.attach(10);  //dedo3

myServo5.attach(11, 576, 2384); //indicador

myServo6.attach(12);  //poegar

 

  } 

  void loop(){ 

   

         int value99 = 0;  //contador de numero de vezes

         int value98 = 0;    //contador de ups

      

       while (value99 < 500) { //Here finds the muscle pattern.

      

      

             if (Serial.available()) {     // if we have new byte from LAN 

           

    if (getOscMsg()==1){        // add it to message while is is not ready

 

    if (byte(oscMsg[1]) != 32){   //para evitar o envio de mensagem vazia

     Serial.print(oscMsg[1]);  // or print the OSC message name 

     Serial.print(" / "); 

     Serial.println(msgVal);     // and vaule 

     int value = msgVal;

 switch(oscMsg[1]){

 

  

    case'a':

 

  if(value > 0 && value <= 180){

  modo = 1;    //variavel para determinar o movimento

          Serial.print("MODO ");

          Serial.println(modo);

  value = 0;

  break;

  }

 

  case'b':

 

  if(value > 0 && value <= 180){

  modo = 2;    //variavel para determinar o movimento

          Serial.print("MODO ");

          Serial.println(modo);

  value = 0;

  break;

  }

 

  case'c':

 

  if(value > 0 && value <= 180){

  modo = 3;    //variavel para determinar o movimento

        Serial.print("MODO ");

          Serial.println(modo);

  value = 0;

  break;

  }

    case'd':

 

  if(value > 0 && value <= 180){

  modo = 4;    //variavel para determinar o movimento

       Serial.print("MODO ");

          Serial.println(modo);

  value = 0;

  break;

  }

      case'e':

 

  if(value > 0 && value <= 180){

  modo = 5;    //variavel para determinar o movimento

       Serial.print("MODO ");

          Serial.println(modo);

  value = 0;

  break;

  }

 

        case'f':

 

  if(value > 0 && value <= 180){

  modo = 6;    //variavel para determinar o movimento

       Serial.print("MODO ");

          Serial.println(modo);

  value = 0;

  break;

  }

 

 }      

    }

    }

           }

      

      

         delay(1);

    

       

        

         ADC_Value = analogRead(CurrentCh);

         value99 = value99+1;

         if(ADC_Value > 400){

           value98 = value98 + 1;  //caso tenha o valor alto, adiciona um no contador

         }

           if(ADC_Value < 250){

           value98 = value98 + 1;  //caso tenha o valor alto, adiciona um no contador

         }

       }

 

      

      

        if (value98 < 25){ //off

        Serial.print("off");    //this is packet header for my application

        Serial.println(ADC_Value);

    

  myServo1.write(90);

  myServo2.write(10);

  myServo3.write(10);

  myServo4.write(10);

  myServo5.write(10);

  myServo6.write(10);

  delay(1);

 

     

        }

        else if (value98 >= 45){   //on

                  Serial.print("on");    //this is packet header for my application

        Serial.println(ADC_Value);

        Serial.println(value98);

       

      if(modo <= 1){ 

          Serial.print("MODO ");

          Serial.println(modo);

  myServo1.write(90);

  myServo2.write(150);   // fechado

  myServo3.write(150);

  myServo4.write(150);

  myServo5.write(150);

  myServo6.write(150);

  delay(1);

      }

           if(modo ==2){

          Serial.print("MODO ");

          Serial.println(modo);

  myServo1.write(90);

  myServo2.write(10);   // pinca

  myServo3.write(10);

  myServo4.write(10);

  myServo5.write(140);

  myServo6.write(150);

  delay(1);

      }

            if(modo ==3){

          Serial.print("MODO ");

          Serial.println(modo);

  myServo1.write(90);

  myServo2.write(150);   // aponta

  myServo3.write(150);

  myServo4.write(150);

  myServo5.write(15);

  myServo6.write(20);

  delay(1);

      }

                  if(modo ==4){

          Serial.print("MODO ");

          Serial.println(modo);

  myServo1.write(90);

  myServo2.write(15);   // Spider

  myServo3.write(150);

  myServo4.write(150);

  myServo5.write(15);

  myServo6.write(20);

  delay(1);

      }

   

             if(modo ==5){

          Serial.print("MODO ");

          Serial.println(modo);

  myServo1.write(90);

  myServo2.write(150);   // Joia

  myServo3.write(150);

  myServo4.write(150);

  myServo5.write(150);

  myServo6.write(20);

  delay(1);

      }

                 if(modo ==6){

          Serial.print("MODO ");

          Serial.println(modo);

  myServo1.write(90);

  myServo2.write(15);   // hang

  myServo3.write(150);

  myServo4.write(150);

  myServo5.write(150);

  myServo6.write(20);

  delay(1);

      }

   

        }

   

   

   

   

   

   

   

   

   

   

   

   

  }

 

 

 

 

 

 

 

 

 

 

 

  // function to process simple OSC message sent by TouchOSC for iphone / ipad

  float getOscMsg(){ 

   inByte=Serial.read();              // read next serial byte 

   if (inByte == 47){                 // if byte = slash it's message start 

    inbyteIndex=0;                  // and we set array pointer to 0 

   } 

   if ((inbyteIndex <= 11) && (inbyteIndex >= 0)){   // is it time to finish or can we start? 

    oscMsg[inbyteIndex]=inByte;            // we add the byte to the array  

    inbyteIndex++;                  // increase array counter 

   }  

   if (inbyteIndex == 11){              // end of the message 

    inbyteIndex=-1;                  // set the pointer to -1 so we stop processing  

    union u_tag {                   // this is array to float conversation routine 

     byte bytes[4];                 // I copied from Arduino.cc forum 

     float buffer; 

    }  

    u; 

    u.bytes[0]=oscMsg[11];              // to decode we have to supply bytes inr everse order 

    u.bytes[1]=oscMsg[10]; 

    u.bytes[2]=oscMsg[9]; 

    u.bytes[3]=oscMsg[8]; 

    msgVal = u.buffer;                // byte array to float 

    return 1;                     // signal we are ready to display value  

   }  

   return 0;                      // in this case the message is not ready yet  

[Alan Hart R & D Robotics]

Just had a look at your video and one of the comments  We would like to put the motors outside of the arm (for disabled people) and Gael langevin comes to meet us in Rennes at the end of June with Nicolas Huchet to explore solutions

 

This person must have been reading my mind, I am thinking of the possibility of the inmoov parts being printed in a way that they can be worn.  Having the motors on the outside in this case would enable a person that needs to have their hand / finger muscles restrengthend in physio. The hand could be worn and would help stop some of the muscle wastage. I have a friend that has nerve damage in her leg and has to have regular sessions where her carers just bend her legs up and down etc etc.

 

 

Alan

 

[gael langevin]

Fantastic Brancante,
I'm glad you have published this. I also heard you got in contact with the FabLab of Rennes, and they were impressed about your work. I beleive this prosthetic system could really be benefic for low budget help.
I have been looking for very small linear actuators to integrate the hand or in the forarm to reduce even more the space used by the servos. The challenge in Rennes is that the person, Nicolas Huchet, is missing the hand, and therefor the forarm isn't needed. So the idea is or to create a socket on the forarm with the actuators, or to totally modify the InMoov hand for to integrate small actuators inside.
This being said, it seems that linear actuators are slow, expensive and not very powerfull and I wonder how this could be processed differently.

[Jackie Huen]

This is a great project. I am a stroke survivor with engineering background, Can you give me some pointer to where I can get information/material regarding how to decode the signal and placement of the pad.

My goal is to make device to retrain the hand function of the stroke survivor.

Thanks for sharing

-Jackie

[gael langevin]

Hi Jackie,

If Brancante doesn't respond here, maybe it is best you get directly in contact with him. He has posted a more recent video with tuto and sketchs.

2013/6/19 Jackie Huen <yfh...@gmail.com>

--
You received this message because you are subscribed to the Google Groups "InMoov" group.
To unsubscribe from this group and stop receiving emails from it, send an email to inmoov+un...@googlegroups.com.
To post to this group, send email to inm...@googlegroups.com.
Visit this group at http://groups.google.com/group/inmoov.
For more options, visit https://groups.google.com/groups/opt_out.
 
 

[Brancante]

Hello Galel,

Sorry for the delay.

The guys from FabLab are doing a great project, I have seen their blog and I am very happy to be of help.

I think integrate actuators in the hand is a great way to solve the problem, but, because of their power and speed, you will probably have to reengineer the fingers, take a look at this article: http://www.sciencedirect.com/science/article/pii/S0957415813000172

I have order some very small servos, maybe if we extend the back of the hand, they could fit in, what do you think?

They are very cheap, so might be worth the test...

Servos:

http://www.hobbyking.com/hobbyking/store/__662__HXT900_9g_1_6kg_12sec_Micro_Servo.html

http://www.hobbyking.com/hobbyking/store/__3715__Hobby_King_S0361_3_6g_45kg_12sec_Micro_Servo.html

[Brancante]

Hello Jackie,

 

As myoelectric is something relative new, I could not find material about it.  My tip is start on the big muscles and attached to the computer to find your patterns and than take down to lower platform like arduino. I have ordered a myoelectric sensor with more channels so soon I will be able to decode more complex movements. For now I am combining the myoelectric stimulation with different modes chosen on a smartphone to be able to do practically any movements.   

 

Have you seen this: http://pranoysaji.wordpress.com/2013/05/30/myo-the-band-that-control-the-computer-by-listening-to-our-muscle-movement/

 

I think might be of help.

[gael langevin]

@Jackie, did you get in contact with him?

@Brancante, yes this is the kind of mechanism that should be thought of, althought creating such small system with our printers isn't realistic. It would be too fragile. I was thinking of making the finger being the mechanism, instead of inserting a mechanism.

The little servos you got aren't expensive and should be nice for testing, I would suggest maybe some servos with metal gears for later. Good progress!

2013/6/20 Brancante <bran...@gmail.com>

--

[Jackie Huen]

Thanks for the quick response. My goal is to show a virtual hand or like you to control a robot. Stroke survivor might be able to send signal to the

muscle, but the signal is weak due to brain injury, or the muscle is weak because of lack of use. Most of the time they don't 'see' the movement

and thinking that they cannot make the hand move, hence use the hand less. This a bad feedback loop. The less frequency to use the hand, the weaker the muscle. 

I hope to build the device to give the survivor a positive feedback, to show them a virtual hand or robot hand to give them visual feedback.

Just wondering if there is any material on the principle of decoding the signal ?

BTW, what is you system configuration ? Ant block diagram for me to learn from.

I hope Myo has a sdk for use to develop on.

Thanks

--
You received this message because you are subscribed to a topic in the Google Groups "InMoov" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/inmoov/vI1sG6WvnQc/unsubscribe.
To unsubscribe from this group and all its topics, send an email to inmoov+un...@googlegroups.com.

[Brancante]

@ Jackie,

 

I see now what you mean! That is a good project. 

Since the muscle might have atrophy, the fine-tuning can be hard, but can be done. 

Everything I use is Arduino based. A Shield from Olimex to get the signal( OLIMEX-EKG/EM) and I a PC Software to see the patterns. The software is ElecGuru. (https://www.olimex.com/Products/Duino/Shields/SHIELD-EKG-EMG/resources/ShieldEkgEmgDemo.zip)

 I started with try and error to manage the first movements, but since you are going to have weaker signals, I recommend you reading this article:

http://www.stanford.edu/~shenoy/GroupPublications/RiveraAlvidrezEtAlIEEEEMBC2010.pdf

 

 

 

 

@ Gael,

 

Yes, that ought be very fragile, thus, not practical! 

 

Have you seen this Linear Servos?

http://www.hobbyking.com/hobbyking/store/__24184__Hobbyking_Ultra_Micro_Linear_Servo_HK_15006_AUS_Warehouse_.html

 

I am developing an Open Source Myoelectric Prosthetic Arm and the size and power consumption of the SparkCore will be decisive.  

Take a look:

 

http://youtu.be/38SnocWQFkM

[Jackie Huen]

Brancante

I may need to duplicate your system and start experimenting on myself.

Thanks for the pointer.

-Jackie

[10beema...@seecs.edu.pk]

Hi,

I have started working on same idea as part of my electrical engineering major project but my area of focus is sEMG feature extraction and recognition. I used this single channel olimex emg shield but I need to get other shield with multiple channels. Despite of lot online search, I haven't been able to find one. Could you tell me what multi-channel EMG shield have you ordered?

[Abrar Satar]

Looks like there wont be much action going on on this thread, but I am doing a Mechanical Engineering Senior Design prosthetic arm and I will be using a mixture of Arduino and INMOOV. 

I think the shield he ordered was:

https://www.olimex.com/Products/Duino/Shields/SHIELD-EKG-EMG/open-source-hardware

and it has the electrode attachment on the bottom of the page.

[10beema...@seecs.edu.pk]

Yes, I have this emg-ekg shield but it is single channel but my requirement is multiple channels. I have heard about multiple channel ads1298 pdk from texas instruments but seems its difficult to be used with arduino through SPI interface. Bracante has above mentioned that he has ordered a different shield with multiple channels, so I wanted to know which shield it is to use it in my project.

[gael langevin]

Hi, I will try to see if Brancante can answer directly. He hasn't been around for a while.

2013/10/9 <10beema...@seecs.edu.pk>

--

You received this message because you are subscribed to the Google Groups "InMoov" group.

To unsubscribe from this group and stop receiving emails from it, send an email to inmoov+un...@googlegroups.com.

[Uhebe UAV]

Guys Awesome Project !

 I want to get started , my 3D printer will be arriving soon , and need to plan , I want to build a right hand and fore arm . to eventually evolve into my end goal which a bionic hand with super strength. this leaves me with a few questions. hope some one can help. with the InMoov hand as is what is the current grip force that the hand can deliver ? equal to human ? more ? less ? other question what is the total cost of manufacturing for a hand and forearm ? i want to start by designing a skin for the hand that has feeling.

[Erin]

Hi Every one, 

I have just completed an Honours thesis on "Controlling a Prosthetic Hand Using Surface EMG". I used the InMoov for my presentation. I did a lot of work on the signals analysis to clean up raw EMG data from the muscle and get an accurate classification of basic motion. I am now moving into Neural networks and AI to look at classifying complex motion using up to 8 EMG channels. 

I have attached a picture of my InMoov here. I will be doing a video in a few weeks and will also post that. Then finally when my honours paper is written I can share with you some of my data. Of course Geal you and all the open source community here have been credited with giving me the ability to build a functioning arm in my own lab. 

There is a lot of potential for this project. I personally have done most of my work in signals analysis and software but this work combined with mechanical improvements can make an accessible, cheap product for potential amputees. 

[on track]

Congratulations

Sent from my iPhone

--
You received this message because you are subscribed to a topic in the Google Groups "InMoov" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/inmoov/vI1sG6WvnQc/unsubscribe.
To unsubscribe from this group and all its topics, send an email to inmoov+un...@googlegroups.com.

To post to this group, send email to inm...@googlegroups.com.
Visit this group at http://groups.google.com/group/inmoov.
For more options, visit https://groups.google.com/groups/opt_out.

<photo1.JPG>

<photo2.JPG>

[Muhammad Abdullah]

Hello Erin,

Congratulations on your successful thesis.

I am working on the same project as my senior design project. Now-a-days I am focusing signal processing using matlab and lab chart.

What surface EMG data acquistion system did you use? I am just wondering is there a possibility of portable data acquisition system for EMG?

--

Muhammad Abdullah,

SEECS, NUST, Islamabad

[Oleg Galtsev]

I read about Bebionic hand and that guys use servos in the hand. I found some information how they build there fingers. It seems very easy(see point (C)): http://www.rehab.research.va.gov/jour/2013/505/images/belter505f02th.jpg
Full description about hands you can find by this link: http://www.rehab.research.va.gov/jour/2013/505/belter505.html

[gael langevin]

@Erin congratulation for your Thesis! I wrote to you some place to congratulate you, but I don't remember where.

@Oleg thanks for the links, I would have liked to get that in july when I was designing the new fingers for InMoov.

There is tonight 25/10/13 a Google Hangout about prosthetic with Adafruit and I will be talking about the new model which, I hope, will help the prosthetic community.

2013/10/25 Oleg Galtsev <ogal...@gmail.com>

I read about Bebionic hand and that guys use servos in the hand. I found some information how they build there fingers. It seems very easy(see point (C)): http://www.rehab.research.va.gov/jour/2013/505/images/belter505f02th.jpg
Full description about hands you can find by this link: http://www.rehab.research.va.gov/jour/2013/505/belter505.html

--

You received this message because you are subscribed to the Google Groups "InMoov" group.

To unsubscribe from this group and stop receiving emails from it, send an email to inmoov+un...@googlegroups.com.

[Brancante]

Hello All,

The Shield I ordered that can combine multiple channels is this: http://www.advancertechnologies.com/p/muscle-sensor-v3.html

But there is no need to combine many, you can work well with one or two. The point is to combine sequences of stimulations to execute different movements. You can also use the wifi on the phone to change the working mode of the prosthetic. 

[gael langevin]

Hi Brancante,

Good to hear from you again. I tried to contact you about Make the world prosthetic, but you come back just a bit too late. :)

As you maybe saw in the last post I have been working on a new hand. That's the one I plan to set on my Beta Tester, Izgi that is her name.

Also, this new hand is not going to work with servos but with motors and potentiometers for feedback.

I'm ready to try to make this whole thing work but I certainly will need help to understand how to get it all operated.

I saw you have made a wiki page about prosthetic project, but I didn't go deep in details yet.

Besides the kit from Advancer, what do you recommand to get this quickly done?

Can you post here the link to your wiki page, I can't find it back right now.

[Joel Hackett]

hi gael
a quick a dirty way to get your hand up and running it to transplant the circuit board out of a servo and wire it up to you're motor and pot, just need to use a servo with the same size motor or bigger and it should be fine. 

[gael langevin]

Thanks guys,

I'm getting prepared for London and can't do a proper answer now, but I have read all your posts and will write again.

2013/11/5 Joel Hackett <jo...@thehackettfamily.org>

[Oleg Galtsev]

Good luck!

среда, 6 ноября 2013 г., 2:48:09 UTC+3 пользователь gael langevin написал:

[gael langevin]

Okay back from London, and a list of 62 emails waiting for answer :)

@brancante, I still didn't order the Advancers sensors, nor the mini Arduino. Next thing to do.

Once I have that I will seak into powering, let me know what you find best.

@jhack, yeah I had tried to hack a servo board with one of my pololu, but it was acting strange. The motor was jittering so I figured maybe it wasn't a good idea at the time. But if you say it is possible, if I respect the size of motor I will try again.

2013/11/6 Oleg Galtsev <ogal...@gmail.com>