11/02/07 Lee Sam

Time in 9am
Time out 4pm

• Continued work on the MAC7111 SPI.
• SPI-A is working when the S_out is hooked up to the S_in (similar to a Loop Back Mode)
• SPI-B not working. SPI-B is the port that had to be soldered connections from. Perhaps problem due to the soldering or other connections.

• Attended meeting from 2:00pm to 3:30pm

11/2/07 Max Wasserman

3:00-5:30

1. Programming

______________________________________


1. I spend my time in lab writing a new MATLAB program to better fit with our objectives. This one takes masses and locations as inputs and tells the user how that additional mass will change the relative inertial properties of the two legs. This change will better help us adapt to the constant addition of new masses to the robot so that we can keep up with how the inertial properties change without having to re-do the testing.

Nov 2 Stephane Constantin

2:00 - 5:30

1) Lab Meeting
2) Test video acquisition matlab code with Panasonic firewire camera

2) The same code that reads video streams from my USB webcam can also read video from the lab's panasonic firewire camera.

11/2/07 Carlos Arango

2:00pm-3:15pm again 4:30-6:15pm

1. Attended weekly lab meeting (1.25 hr)

2. Initiated calibration procedure for ankle sensors (20 min)

3. Switched to helping Ying Yi with her sensor calibration (1.25 hr)


----------------------------------------

1. Self-explanatory

2. While Ying Yi was setting up the magnet in the hip sensor, I took a couple of readings with a connected ankle sensor to get the value of the gain and offset we will need to calibrate the sensors. I have implemented the value of the gain into the code, but I have to take further readings to make sure the offset is correct for that sensor.

3. Because Ying Yi's project has higher priority, I have switched to helping her calibrate her sensor, as we can't both use the computer at the same time.

11/2/07 Ullmann Kevin

3:30 - 4:30 (1 hrs)

1. Discussed options for video with Stephane (.5 hrs)
2. Wrote user manual for graphics program (.5 hrs)

1. We discussed issues that Stephane has been facing in his attempt to create the video component of the graphics program. Mainly the issue of ginput waiting for a key press, thus halting execution of the program. We believe that this problem can be solved using threading, or perhaps would be easier to work with in LabView.

2. I created a concise user manual for the current version of the graphics program, so that anyone may use the program with ease.

11/02/07 Hajime Furukawa

12:00 -6:00

1) Built 2 sensor boards. (6h)

-------------------------------------------------------

1) Added Voltage regulator, connectors, Muxes, OP amps, and all other components to two sensor boards.

11/1/07 Kevin Ullmann

11:00 AM - 1:00 PM (2 hrs)

1) Created initialization process (1 hr)
2) Fixed help display (10 minutes)
3) Testing and research (50 minutes)

1) I created an initialization process to streamline the task of running the graphics program. Now the user simply has to run a script and will be prompted for the necessary inputs, including the data file and the desired variables.

2) I altered the help display message to be more readable and user friendly.

3) I tested the graphics program and corrected some minor bugs, and I researched the task of aquiring the mouse position for use with the vertical cursor. I am still working on this issue.

11/01/07 Yingyi Tan

11.00 am - 1.45pm, 4.00pm -6.00pm

1. Moved motor in left hip box 2 mm to the left. (1.5h)
2. Milled away material on clamps holding leg to box (0.25h)
3. Helped Jason enlarge the hole for the cable in the left hip box (0.25h)
4. Helped Jason drill and tap hole to hold the bottom of the left hip box (0.25h)
5. Drilled holes to hold the hip encolder spring to the robot (0.25h)
6. Fit everything together. (0.25h)
7. Put electronics and clamps back into hip box (1h)
8. Labelled the extra allen wrenches (1h)

------------------------
1. After putting the sensor board with the pins into the leg, we found that there was no clearance between the motor and the pins for the spring. Decided to move the motor to the left. however, the cables were rubbing against the clamp holding the leg to the box. Had to remove the clamp which was obstructed by the metal sheet holding the microcontroller board to the robot. Removed the board then removed the clamps.

2. Used the end mill to remove the material above the top hole in the clamps.

3. The cable was also unable to move further left as the hole on the bottom of the left hip box was not wide enough. Jason enlarged this hole with a dremel tool. I had to vacuum the robot while he drilled to prevent the metal shreds from touching the electronics.

4. Enlarging the hole in the above step removed the weld that was holding the bottom of the box to the box itself. Had to drill and tap a hole through the bottom and the box to hold the two pieces together. Once again, I helped Jason by using the vaccuum cleaner to prevent the shards from touching the electronics. However, the process took much longer than expected as there was insufficient room to put the hand drill. Had to improversie and make our own dril bit extender with screwdrivers.

5. Used the hand drill to drill two holes in the robot to hold the hip encolder spring to the box.

6. Put everything together (hip encolder housing, spring, sensor board and pins)

7. Screwed the clamps back to the box with loctide. Screwed the microcontroller boards back to the robot.

8. Measured the unlabelled allan wrenches and labelled them using heat shrink. The wrenches were color coded depending on whether they were english or metric sizes.

11/1/07 Carlos Arango

12:10-1:10 pm and again 2:45-5:10 pm

0. Looked for assembled sensor cables (15 min)

1. Routed 1 sensor cable up the leg to test cable mounting bracket (15 min)

2. Filed mounting brackets to ensure they are safe when cable is attached (30 min)

3. Worked with Jason debugging code for the satellite sensor board (1.5 hrs)

4. Read through sensor input code to gain understanding of its workings (25 min)

---------------------------------------------------------------------------------------

0. The individual who constructed the sensor cables forgot to mention where he had stored them upon completion. Turns out they were in his work bin

1. Attached the sensor cable to the mounting bracket and leg of the robot with cable ties. I made sure to be liberal on the cable tie use to allow others to work on the legs after my cable was mounted.

2. Jason pointed out that the mounting brackets still have sharp edges. I carefully went through and filed those off. Since the brackets were already mounted, this took a lot longer since care was needed

3. I was invited to help Jason debug code that allows the sensor board to communicate with the main processor board.

4. My next task will be to test out the ankle encoders with the code Jason has written. To do this I felt it was necessary I first understand the inner workings of this code

11/1/07 Gregory Falco

1:30-4:30PM

1. Created website for publicity
2. Fixed up original copies

-------------------------------

1. Created a website using word that links all publicity files to one page. I am still creating links from the general file to the article titles.
2. Wrote down page numbers on all original publicity and dog-eared pages.

10/30/07 Gregory Falco

12:00PM-5:00PM

1. MSDS (4.5 hrs)
2. Unscrewed Wooden Structures (.5 hrs)

_________________
1. Finished Material Safety Data Sheets for the lab and machine shop and created a draft for website on MSDS. Now there is a physical and virtual set of MSDS for both the lab and machine shop.

2. I dismantled one of the wood structures that was outside of the professor's office and put the scrap in the machine shop.

10/31/07 Lee Sam

Time in 9am
Time out 2pm

1)Rewire SPI port for the MAC7111 - wires got snaged un handling
2)Wires super-glued and soldered in place to prevent repeat of problem.
3)Tested outputs from SPI ports on oscilloscope and outputs appear ok.

10/31/07 John Buzzi

4:30-7:30

1) Successully getting microsoft project working in Upson (30 min)

2)Use Project to develop a network diagram and GANTT Chart for the Walker (2 hr 30 min)

-------------------------------------------------------------------------

1) I found that the upson computer lab has Microsoft Project on the computers. The purpose of Microsoft Project is to help visualize and organize a project plan, in our case we want to see what needs to be done before the Walker is complete. However the versions installed in the lab aren't working properly. It took about 30 minutes of switching computers and clicking on error messages to get the software to open somewhat properly. Then once i completed a few steps and clicked save...it deleted everything i had done.

2) Once project was up again and working slightly better than before I began work on the GANTT chart. A GANTT chart is basically a timeline of tasks in a project going from start to finish. Data input for each task includes, task name, names of people working on it, which tasks must be completed before the task at hand is completed, and the duration of the task at hand. I used Jason's power point slides to gather the necessary information. Also once all the data was input a "network diagram" is automatically generated which is a better way of seeing dependencies, however much formatting had to be done in order to make it of any use.

10/31/07 Max Wasserman

1:40 pm - 5:10 pm

1. MATLAB Testing

2. Inertial Property Testing

-----------------------------------------------------------

1. I continued using the MATLAB Program to determine the optimal points on which to add more mass. However, that data will have to be altered based on the values obtained from how I spent the rest of my lab time.
1 hour

2. Rohit and I re-tested some of the inertial properties of the robot, as since several masses had been added since our last test. We only tested the inner legs as they are where most of the new added mass is located. We did the swinging test to determine moment of inertia about the him and determined the angle of the center of mass and the value of m*r.
2 hours, 30 minutes

10/31 Stephane Constantin

10:30 to 2:30 at home

1. Reading video for graphing tool

I continued to explore how to capture video frames from a webcam and save the video framesto an AVI file using Matlab. I pretty much have this part working now. I also started to explore how to read that AVI file and display desired frames accordingly. This feature will be used to target which frame we want to seeaccording to at what time stamp the graph vertical cursor is.

10/29/07 Lee Sam

Time in 9am
Time out 2pm

• Tried to get the 2 SPI on the MAC7111 to communicate with one another, currently still unsuccessful.

• Looked at the signals coming out of the 4 leads on the SPI master and verified correct.
  
• Tested soldered connection and soldered pins - the correct pins were soldered and the connection between the pin and the soldered wire-warp. No problems found in this area.
• When a test code is run, the program steps through the portion of the code that sends the data - and recognises it as sent (the transfer counter is incremented), but the other SPI on the receiving end does not receive it.
  
• Currently unsure of what problem is preventing the 2 SPI from communicating, will continue to troubleshoot.

10/30/07

5.30 - 7.00pm

1. Re-machined the conical pin (1.5h)

--------------

1. The brass conical pin was too large to fit into the hole on the sensor board. I originally filed it down but this resulted in the pin no longer being circular. Hence a new pin was machined with the lathe.

10/30 Rohit Hippalgaonkar

Hours : 10.45 - 11.45 am
2 pm to 4 pm

1. More refinements to results from code. (2.5 hours)
2. Finding appropriate blocks of scrap etc that can be added (0.5 hours!)
--------------------------------------------------------------------------------------------------
1.
Discussed results with Pranav and Jason. We were able to further reduce the sum of masses to be added to 0.65 kg. Besides, this time the masses were tried in locations that could actually be used on the robot (for e.g. a battery near the joining bar on the inner leg, a controller in a plastic box just below the aluminium box on the outer leg)

Example case:
-----------------
If we place the mass I (ref. in last post) as far down below as 75cm from hip axis on inner leg, with mass III located 6cm out at the hip axis (same as last post), and mass II just below the box on the outer leg (12 cm) and about 5 cm out from the plane of the robot the masses are:

I. 0.29 kg, II. 0.165 kg, III. 0.17kg.

Note that mass II must be split equally on the 2 outer legs.


2.
This took a while, candidate pieces of scrap found for I and III and one flimsy piece of wood for mass II found after much searching!

10/29/07 Yingyi Tan

5.30pm - 8.30pm

1. Drilled a larger hole in the hip shaft (0.5h)
2. Sandpapered hip shaft (0.25h)
3. Fit assembly together on robot(0.25h)
4. Re-machined the hip spring and brass pin. (2h)

-----------
1. In order to create more wobble room for the hip insert, we drilled a larger hole where the top portion of the threaded hole used to be. This allowed the hip insert to fit into the hip shaft.

2. Checked that the hip insert was flushed with the hip shaft. To ensure that the hip insert was completely smooth at the contact point with the hip encolder housing, I used the microscope and sandpaper to file out any parts that were sticking out.

3. Put the middle leg back on the robot and fit the assembly together on the robot.

4. The hip spring was too narrow. It was difficult to put the screws in as the motor was blocking access to the screws. The old hip spring had two holes for the pins connecting the sensor board. The new hip spring only had one hole. The screw holes were made into slots to allow for horizontal movement. The original pins were made of aluminum. Jason commented that this might cause the aluminum pin to bind with the aluminum hip spring. A more suitable material would be brass. A new conical pin made of brass was machined using the lathe.

10/29 Rohit Hippalgaonkar

Worked from home - 4pm to 8pm

1. More testing on code - drew some conclusions (2 hours)
2. Further work on program to calculate effect of solid friction (~1 hr)
3. Updating blog (45 mins)
-------------------------------------------------------------------------------------------------
1.

Problem
Of the 3 mass properties to be equalized for the 2 legs currently, the inertias (about hip) and m*y have large differences (~1600 kg.cm^2 and ~20 kg.cm), compared to the differences in m*z (~ 2 kg.cm). Hence either the input distance(s) to the program must be large (to get reasonably low masses to be added ), or we should expect high masses to be returned by the program for location inputs that are all close to the hip axis.


Approach
I wanted to keep it simple -> The analysis is for the case of 3 masses to be added. We would have a simple (linear) system with 3 equations and 3 unknowns. Importantly, we always get the new differences in the 3 mass properties to be exactly equal to zero.

I) Now, to equalise 'm*z' (which is low) I only added a mass on the outer leg that is off-plane but has no y-moment. This is the only mass with a z-moment.
II) Similarly to equalize the large difference in inertias we will have a second mass far away from the hip-axis but without a z-moment.
III) I arbitrarily chose a y-location for the the third mass but again this has no z-moment (could be on either leg, we just get negative values with the same magnitude).

This simplicity in the location of masses helps in drawing conclusions, I feel. Also this may help in actually adding the masses as well.

Results
If the second mass is kept at distances less than 35 cm (below hip axis), we get large values for sum of masses (6.1 kg and higher). With y = 40 cm for the second mass and varying location of the third mass (10,15, 20, 25, 30) the sum of masses is 3.2 kg. Simliarly y = 45 cm, y = 50 cm, y = 55 cm were tried and this further reduces the sum to as low as 1.7kg (y = 55cm). For all the above locations, the sum was least (and more or less constant) when the third mass was kept between 20 - 25 cm.


Consequences
* large y-moment of second mass about the hip axis leading to :
a) greater influence on the motion of the outer leg when it is the stance leg.
b) helps in swing of inner leg for first half of the motion (until vertical), and opposes motion for the second half.
c) reduces opposing moment when the inner leg is stance leg, as the second mas is closer to the ground.

At any rate I feel adding the second mass below the hip-axis is better than adding it above it - this would really increase opposing torque when the inner leg is stance leg.

Possible Refinements
Further reduction in sum of masses could possibly be achieved by adding a z-distance to the second and/or third masses or a y-distance for the 1st mass.


2. Program still not running, apparently MATLAB has problems with Coloumbic friction models.

10/29/2007 Kevin Ullmann

3:30 - 5:00 PM (1.5 hrs)

1. Fixing small problems (1.5 hrs)

1. Today I worked on touch ups to the graphics program including removing the "y=" from the value labels, adding a "Press H for help" line to the top of the window, fixing small bugs, and adding the variable numbers to the y labels.

10/29/07 Gregory Falco

12:30PM-5:00PM

1. MSDS (4 hrs)
2. Rectifying Dealtree Refund (.25 hours)
3. Blog maintenance(.25 hrs)

---------------

1. Created Labels for each chemical and began attaching them to each hazardous material we have in the lab.
2. Found old e-mail confirming camera (dealtree) transaction that had not been successfully sent to Ms. Moss. The account had never been charged due to this so i resent all necessary e-mails and hopefully cleared up the potential issue.
3. Fixed up blog so that everyone's entries are uniform

10/29/07 Avtar Khalsa

12:30Pm to 2:45PM

1. Used Code warrior simulator to make sure that we could break code into smaller files for projects. Did this with sample code -hour and a half

2. Looked at Jason's attempt to break up actual ranger code to try and see what the problem was. -45 minutes


1. I used a code warrior sample code for our microprocessor and broke it up into three different files. It needed a header file, and implementation file, and a main file. Then I used the simulator to make sure that the code still worked. It seemed to continue working correctly from which, I concluded that it was definitely possible to break up the code like we planned.

2. Jason had previously attempted to break the code up into many files but was unsuccessful. I spent about 45 minutes scanning over what he had done trying to find what the problem was. It is very difficult though because there isn't really any way to test it until the robot is put back together.

10/29 Stephane Constantin

10:00 - 12:00

1) Made connectors for ankle sensors to satellites
2) CAN tranceiver adapter documentation

10/28 Rohit Hippalgaonkar (for 10/25 through 10/28)

Hours : 10/25 - Thursday 10 .15 am to 11.15 am
10/26 - Friday 10.30 am to 12.15 pm
10/28 - Sunday 10 pm to 1am

1) Mainly, further testing on the code. (3 hours)

2) started writing programs to see the effect of (solid) friction on i) natural frequency, ii) resonant frequency of each leg. (.5 hours)

3) Helped out Stephane with wiring for a bit and listened in on the theoretical issues with James' project when Prof Ruina was talkig about it. (~ 45 mins overall)

4) some time spent discussing TAM 570 HW with David Blocher next door on Thursday, and Controls HW with Pranav (1.5 hrs)
------------------------------------------------------------------------------------------------

1) Seems best to add three masses on the robot - i) one mass off-plane on the outer leg at the same height as the hip, ii) one mass at a large distance below the hip on the inner leg and the iii) third mass at an intermediate distance to be placed on the inner leg (the last 2 masses both being in the plane of the leg).

Minimum value for total mass to be added seems to vary between 2.15 kg to 3.2kg depending on where we place a mass on the inner leg (tried for a range of 40 cm -50cm which has least sums). Details of analysis and approach used to arrive at a good solution will follow in next post.

Will actually add masses Tuesday.

2) MATLAB gets stuck when trying to solve with solid friction. Running this program is important to our project, since it will give us an idea of the accuracy of the approximations made while calculating inertias from experimental frequencies observed (especially in the case of outer leg, as it had high friction while swinging - coming to a stop in just 5 cycles).

10/28/07 Yingyi Tan

3:30 pm -8:00 pm

1. Continued filing the hip insert to fit into the hole in the hip shaft. (1h)
2. Discussed ways to make the insert fit into the hole without anymore filing (0.5h)
3. Made a new hip insert for the hole (2h)
4. Tried to fit the new insert into the hole (0.5h)
5. Made the screwdriver slots in the insert (0.5h)

-------------------
1. Used the CNC machine to grip the insert while sand-papering the outside.

2. Jason commented that sandpapering the insert any further will cause the hip insert to wobble and the whole assembly will no longer be centered. We discussed ways to make the insert fit into the hole. This included shaving off the threads, hamering the insert, cutting the threaded portion off. Concluded that the best method may be to make another hip insert. This time ensuring the insert does not bend.

3. Made a new hip insert using the lathe machine.

4. The new insert was still unable to fit into the hole. We tried hammering on the insert, hoping to bend it to fit the hole. Concluded that the pre-existing threaded-hole may not have been centered. We then tried shaving off the threaded portion and hand-drilling a slightly larger hole in the shaft. However, the insert could not fit in all the way. We decided to hold the robot in the milling vice and drill using the mill tomorrow.

5. Made the slots in the hip insert for the screw driver using the mill. Had to screw the insert into an aluminum piece to secure it tightly.