16.400 Human Factors Engineering
CDIO Lab#3: Manual Control Lab

Introduction

The MCL2.EXE tracking software is used to present the basic principles of human control and tracking. Using a mouse as a form of input, various factors in tracking-control will be examined in this experiment.

This lab will introduce three of the more common control elements: proportional (0th order), integral (1st Order), and a double integral (2nd order) control. You should all be somewhat familiar with these controls from your controls course. If not, don't worry, hopefully by the end of the lab you will have a good idea of what is going on!

You are welcome to work with others, so long as you each do every part of the lab.

The Actual Lab

The general system you will be looking at is:

Part 1: The Step Response

NOTE: Instructions on installing and running the software follow after the lab description. PLEASE look at them before starting the lab. When you start up the MCL2.EXE program, you will be asked for a configuration file, choose either HIGH.CFG or LOW.CFG. Configuration file does not affect the step response.

We will use the unit step response to familiarize ourselves with the program and how to control the cursor. In the unit step exercise we will be trying to move the cursor into the target. I recommend that you play with the program and unit step response section a little bit before starting to actually perform the lab.

When you feel comfortable with the program you can perform the following cases. For each of the cases, note the movement time. The following investigations can be performed and documented:

• Perform a general run for each of the controllers (0th, 1st, and 2nd order response) -- set the open loop gain to 2 for each of the initial cases. For 1st order response, start with a damping factor of 1.
  
  
  • Note the qualitative workload for each (i.e. On a scale of 1-10, what is the perceived workload?). Do this with all options set to the default (except for time out=45 sec).
    
    
  • Get a screen capture for your report for each of your basic controller responses, if possible. What effect does each of the controllers have on the system dynamics & output (overshoot etc.)? Why? Note the response of each control type. How is each controller different? Maybe a sketch of the response to a unit step input could be beneficial in convincing yourself. Did you notice any unexpected results from your trial runs? Why do you think they may have occurred?
    
    
• Briefly investigate the effects of increasing and decreasing the gain. Try several different gains. What effect does this have?
  
  
• Experiment with various time delays (from 0.1 sec - 3.0 sec). Give a qualitative rating (out of 10) of the workload as time delay increases. Provide some sense of the critical time delay for each of the controllers (where does system instability prevail). In what applications would time delay be a concern -- give some examples where time delay is inevitable, and we have to work around it. Suggest a way of getting around time delay problems.
  
  
• Investigate briefly the effects of adding damping in the first and second order systems. Briefly describe what is theoretically happening as we introduce these features? Did you find a significant difference?
  
  
• If you are doing this as a group, you can also do a secondary task assessment (such as turning a flashlight on while your friend is on the control, and seeing how long it takes them to tap the light or asking them to read the time from a nearby clock (and seeing how accurately and efficiently they do it).
  
  

For each of the cases, how does the movement time for each of the cases relate to the perceived workload? Plot a quick graph to see if your scale has any relationship to the movement time in the unit step case.

Part 2: Low & High frequency Random Disturbance

You will perform a High and low frequency disturbance control for each of the controls. Here we will look at tracking a low frequency random disturbance, and a high frequency random disturbance. The low frequency disturbance is set up in LOW.CFG, and the high frequency disturbance is in HIGH.CFG. You will have to load the case you are working on before the frequency is changed (so once you are done the low frequency cases, load up the high frequency ones by restarting the program).

To start this part, you will have to go back to the main menu, and select compensatory tracking (try a run with pursuit tracking, note the differences between pursuit and compensatory, and perform the rest of the lab for just the COMPENSATORY TRACKING case).

NOTE ABOUT SETTINGS: Change the sample time in the general configuration to 120secs (1st menu level after main screen, on the way into tracking). If you get a memory error, try a lower time.

You will need to run a case for each of the base controllers at each of the frequencies. Here you will be looking at:

1. The error in tracking (RMS). Here you will want to compare the error value for each of the controls, for each of the frequencies. From the errors, what can be concluded? Also what percentage of time on target do you get? Is this expected?
   
   
2. What effect does the frequency of the disturbance have? How well can you control/compensate for higher frequency oscillations? From your results what do you feel is the highest frequency you can compensate for?
   
   
3. The human transfer function & the plant transfer function in terms of Bode plots (A print out of the bode plots will be useful). Discuss to what degree the human counteracts the effects of the plant control function. How does the human transfer function change for the different cases? Why does it act like this?
   
   
4. Workload assessment: for each of the cases, what is the perceived mental workload. Once again if you are able, do a secondary task measurement (only for groups).
   
   
5. Try running your system with a time delay for each of the controls. Put in half the critical value, note results, and workload. Is your system stable for each of the controllers? If it is try 3/4 the critical value, if not try 1/4 the value for each of the cases. What conclusions can you draw from this?
   
   
6. The second order system is virtually impossible to control. From part 1, what do you think may help 'stabilize the system'? Try to implement it if possible.
   
   
7. If you have some extra time and you have had fun chasing your cursor across your screen, and you found the lab was easy, try to investigate the effects of prediction/quickening on a second order system. Set your position gain to 1 and vary the velocity gain. Try several cases where you are trying to determine what range of K values make the plant more stable. I will provide more information on this if you want to give it a try. Just let me know by email, or in class.
   
   

Reporting Requirements

Due date for the report is Novebmer 15th.

Some comments should be made on the three different types of controls. These should be based on observations from the unit step response and the frequency response. A brief summary of the applicability and the suitability of each of these controls should be made. List down some places where we see everyday application of each of these controllers.

Use the guides in the lab requirements to determine what should be included in the report. Be concise with your conclusions and results (unlike this document). This lab is to help you gain an appreciation of the different types of controllers, and the various functions associated with them, so please do not write an essay about why your dog has poor tracking performance with a mouse, etc. Graphical results with a brief explanation will be adequate for the majority of the results -- relevant, short and sweet.

If you have any questions or comments, feel free to email me at tatsuki@mit.edu. I would like to hear back from you if you thought this lab was too long, short, easy, hard, if you thought it was useful, what you think could be improved, etc. It will benefit myself as well as future students in the course!! The comments will be useful in evaluating this lab, and how I should modify the other labs, so that you can get what you want out of them!!!

Installing the Software

Before you start installing the software:

1. Make sure you are using a machine that will run DOS programs.
   
   
2. Remember that this software is for your use only. Please do not distribute this software to other people outside the class.
   
   
3. If you do not have access to a DOS machine please contact tatsuki@mit.edu for more information on how you can gain access to a computer. The computers in the lobby in front of the Aero/Astro library run Windows98.
   
   
4. Clean your mouse. Take out the ball and clean the wheels that pick up the ball's motion.
   
   

Downloading the Software:

1. Create a directory for the MCL software (short name without space is recommended e.g. C:\MCL).
   
   
2. Go to the Lab#3 web page ( http://web.mit.edu/16.400/www/mcl/), and right-click on the files to download.
   
   
3. Navigate to the downloaded files, and doble-click on MCL2.EXE. This should bring up a DOS window. If this doesn't work start a DOS window from the start menu, change to the directory, and run MCL2.EXE.
   
   
4. Hitting [Alt]+[Enter] toggles the full-screen mode.
   
   

Program Familiarization

Now that the software is running the hard part is over. The fun is now beginning. There will be several tasks that you will perform while using this software. As you perform the various parts keep notes and screen captures (this is described later) so that the report will be easy. You may work in groups or alone, however, everybody must perform the control exercises!!

1. Start the software, by the method above. You should come to a screen that looks like:
   
   
   
   
2. You will want to use the down arrow key to highlight the case you wish to test. We will look at the file LOW.CFG first. Highlight the file and press [enter]. The main screen as follows should appear:
   
   
   
   
3. This is the main screen. It is where the initial selection of cases is given. We will look at Step response to begin with. In Part 2 of the lab we will look at the Compensatory and Pursuit tracking. By pressing [1] we go to the next screen (step response menu).
   
   
4. Here we have the option to change the configuration, or to run the case. We will take the opportunity to change the configuration at this time, by selecting <1>. Now it should bring up a new screen. Change the trial timeout option to 45sec. When you have done this return to the previous screen.
   
   
5. We are now ready to run the trials! By pressing [2] at the step response menu we come to a screen that looks like:
   
   
   
   
6. This is the screen that sets the parameters. Press [2] to change the dynamics of the system, when required.
   
   
7. Various changes will be made as we go along, but this is the general way to set up for the first part of the lab.
   
   

NOTE: IF THE COMPUTER CRASHES, OR YOU MESSED SOMETHING UP, JUST CLOSE THE SOFTWARE, AND TRY AGAIN. IF THE HIGH.CFG & LOW.CFG FILES GET MESSED UP, DOWNLOAD A NEW SET!!

SCREEN CAPTURES -- for output of results

For those using windows, and want to get output similar to what is shown above follow the following steps:

1. Run the program in a window in windows.
   
   
2. Press [PrtSc] or [printscreen] button (top RHS of keyboard) when you have a screen you want to copy.
   
   
3. Open paint or a similar program and press [ctrl]-[v], to paste it.
   
   

REMEMBER to invert the colors before printing or copying to MS-word.