Receiving Info From the Servo Motors

      Since we successfully sent commands to the controller using the sliders, our next endeavour was to get feedback on the actual position of the servos as they move, using the commands Q & QP. The format of the Q command is "#(servo) Q", and it will send back an ASCII '+' or '.'. The plus means that the servo is currently in motion, while the period means the servo is stopped. The QP command takes the form of "QP(servo)", and will return the pulse width that the servo has been delivered. Using these 2 commands, we can track the servo as it moves.


      We added a text box to the right of slider 4 in our control window and modified our code to have the program show the current pulse width as the slider and the servo both move.
     Our first attempt at coding this was to add a subroutine called “getPulseWidth” that sent the SCC-32 the QP command for servo 4, then read the response and put it in the text box. This operated on roughly the same lines as how we sent sent the move commands to the arm, and then it used the 'read' command to get the signal that was sent back. Because the SSC-32 can sometimes take a few milliseconds to respond, we included a timer that pauses the program for 6 milliseconds to account for that. The program then converts the response to to an integer, and places it in the text box.

      This worked, but we wanted something that would constantly display the pulse width as it changed, not just once when the slider was moved. To do this, we used a loop to call “getPulseWidth” multiple times, constantly updating the text box. It kept getting the value but introduced a few other problems. Since that the loop was constantly running, it became somewhat of a power-hog, and was using processing power when we didn't need it to. This caused large amounts of lag, slowing down our program.

          To fix this,we used a special timer that would trigger every 5th of a second and check the pulse width. This worked perfectly but the timer was still running. We fixed this with the “Q” command, so that the timer when the servo stopped moving. It works by sending the Q command to the arm, reading the response, and if it's a period, meaning that the arm has stopped, it kills the timer.

      We also did one last check of the pulse when the arm had stopped to make sure we got a final position value. This ensured that the position in the text box was correct. We also had to add a couple of lines to kill the timer before we started it just to make sure that the timer was completely reset.




Here's a video showing the final program with the pulse value being updated as the servo moves:

Robot Arm: Writing The Controller

     Since my understanding of programming isn't overly advanced yet, the best approach to building a program for the arm would be to find a program that has a similar function, and modifying it for my needs. After browsing the apple forums, we found a program that someone had developed that could output signals from the serial port. 

       Using a programming environment called XCode, we opened up the program and examined it to figure out how it worked. We then stripped out all of the code we didn't need, leaving us with a solid foundation that we could build upon.

     


       From there, we built a simple interface consisting of six sliders that controlled each of the six servos. The program had to be written in a language called Objective C, which has some similarities to the Java programming language we used in my Grade 11 Programming course, so it wasn't that hard for me to understand.

       Here is a look at a portion of the code that handles the interface. The very first line basically waits for any of the sliders to be changed, and when one is, it activates the code within the curly brackets. The next line creates the format for the signal that will be sent to the robot(the red text). Normally, the signal that is sent to the robot is "#(servo) P(desired servo position) S(speed)". For instance, if you wanted to send the shoulder to position 1700 at a low speed, you'd type "#1 P1700 S100" and then hit enter. So in our program, you'll notice the same format, except the servo number and servo position have been replaced by a "%d", which is used in Xcode as a substitution, so you can get the program to replace that "%d" with any set of symbols, letters, or numbers. 

        The next few lines get somewhat complicated, but all they really do is take the decimal data received from the slider and convert it to ASCII, so the arm understands it. Finally, the last line outputs the ASCII command through the serial port.

Robot Arm: Testing the Computer

     To run the robot arm with a computer, we first needed to make sure the computer could send and receive signals. Since the Macintosh does not have a physical serial port, we used a USB to serial cable that we connected to the physical serial port on an older PC. Then, using terminal emulation programs, which allow you to send and receive signals out the serial port. We set both of the programs to accept ASCII characters, as that is what the robot arm uses, then successfully transmitted a signal to the PC using the Mac, with the same speed and conditions required by the robot arm(115.2kb/s, 8 bits no parity).

Here, you can see that the message that was written on the mac was successfully delivered to the PC using the terminal programs.

     We later used the same method to test if our simple program to control the arm was working. It was better to use the PC to test the program, rather than the arm itself, because we can actually see the signals our program is sending. We used RealTerm on the PC ( download at http://realterm.sourceforge.net ) and CoolTerm on the Mac ( download at http://freeware.the-meiers.org/#CoolTerm ). Both are free and work really well.