Anna's Blog for Summer 2007

Week 9

August 24, 2007

• Gave presentation at group meeting.
• Worked on SURI presentation and documentation.

August 23, 2007

• Took force sensing data while SB2 climbed in order to determine the range of forces to be expected. SB2 climbed well with little more than a thorough cleaning. Used matlab real_time_graph to chart the A2D? clicks as SB2 climbed. Found that +/- 10 A2D? clicks (~.13 N) described the range, which is well within the linear region of the curves that I obtained for the flextures.
• Worked on presentation for lab meeting tomorrow -- thought about the graphs so that I can explain some of the phenomena present. One particular graph, where I compare the stiffnesses of the various possible compliance methods, was problematic. The curve of the rubber band line suggests that the force is constant as the angle is increased -- and since F is constant and x is increasing, k must be non constant and decreasing with additional force. Still difficult to get my head around.

August 22, 2007

• Finished data manipulation, although still considering the analysis.
• Worked on updating the CompliantWing page.
• Started on my presentation for Friday's group meeting.

August 21, 2007

• Continued analysis of curve data. Found that 2nd order polynomials describe the curves better than linear models -- however, there are also non-linear outliers when the force exceeds .5 N or so. Hysteresis was apparent -- Sal's recommendation was to remove it and to adjust for it in the force domain (within the software).
• Worked on updating the CompliantWing page with graphs of the data.

August 20, 2007

• Completed calibration testing with SB2.
• Began analysis of curve data -- making relationships between voltage and force for each sensor (there was quite a lot more variability than expected among the sensors).

Week 8

August 17, 2007

• Continued the work from below.
• I was having a problem where the scale would get stuck at zero for many subsequent clicks. I solved this problem by returning to zero clicks of offset and then quickly back to the value I desired. This worked reasonably well.

August 16, 2007

• Began collecting data for calibration. The force snapshot provided by the matlab code returns the number of ADD clicks. I'm relating this to the actual force felt by using the number of clicks (up and down) and the scale reading. My test range is -30 to 50.
• So far the data is relatively linear and repeatable. There is some variance in the force snapshots, even when at the same scale reading -- so I have decided to take three rapid snapshots and average them for my curves. I'm doing two curves for each leg, to verify that I have observed a characteristic relationship.

August 15, 2007

• Started SURI/Group Meeting presentation
• Finished verifying that all of the sensors are connected and their feedback can be read in real_time_graph. There were a couple whose connections were not soldered completed at the receiving piece for the four-prong connector. Sal helped me track these down and fixed them.
• Began design of calibration method. Fabricated temporary stand for SB2 and acquired all of the materials. Updated matlab code for force feedback from the wings as well as the stroke sensors. Will begin data collection early tomorrow.

August 14, 2007

• Worked with Sal to debug the front right leg on SB2 to incorporate feedback from both the wing and the stroke sensors. Found that there was a broken connection which was causing the wing sensor to read a maxed out value of 5 volts. Once that was fixed, both worked in real-time.
• Finished soldering rear two sensors.
• Installed the three remaining magnet sets and sensors.

August 13, 2007

• Pre-tested hall effect sensors and selected four to use on SB2.
• Connected wire tapes to 4-prong connectors and soldered to sensors.
• Installed one set of magnets using tape and place technique that did not require disassembling SB2, and glued in one sensor after insulating the soldered connections with hot glue. Will test this sensor before installing the rest.

Week 7

• Learned about behavioral control from Sal -- including gaited and state control. Apparently, it is not similar at all to E105 controls, but it doesn't seem too bad. Referred to Sal's control presentation on the twiki and some papers by Raibert.
• For the time during Darpatech, I have been collecting, reading, and digesting this material. Some has proven to be interesting and possibly informative, and they're listed here: AnnaLiteratureReview.

Week 6

-- Out of Town --

Week 5

July 27, 2007

• Entered final data sets into Excel to finish graphs for compliance testing and updated the twiki page accordingly, along with some results/interpretations: CompliantWing.

July 26, 2007

• Used test setup to test the 3 mm and rubber band wings, as well as the 3 mm back wings, which were made more recently and in a different batch that had a semi-tacky surface cure. Noticed that there was in fact a definite spring constant difference.
• Worked on graphing the data and making comparisons.

July 25, 2007

• Assisted Barrett in trying to get SB2 to climb with its new parts.
• Constructed an apparatus to test the loading on the wing flexture -- uses a servo board that Barrett programmed to allow me to step through various angles and record the force measured. I then am able to convert these to torques at the flexture and to compare between the 4 mm, 3 mm, and rubber band wings.
• Took initial tests of the 4 mm flexture and graphed the data to see what it looked like. It's relatively linear and consistent.

July 24, 2007

• Cut toe pads for SB2's new two-toed feet and glued them on.
• Worked with Barrett on setting the new home positions. SB definitely sticks to the glass, but some more adjustments will be necessary before it climbs.

July 23, 2007

• Pried out remaining back wings. The soft polymer did cure this time, however, it was still slightly tacky on the surface. This didn't seem to effect the strength of the flexture or cause delamination, although it's possible that a difference between these tacky back wings and the previously made front wings will be detectable in testing.
• Constructed back wing components
• Removed SB2's feet and wings to replace with my wings and Barrett's feet.

Week 4

July 20, 2007

• Machined soft part mold for back wings, and poured the polymer.
• Brainstormed test apparatus for wing-testing.

July 19, 2007

• SB dry-run day -- suffered some problems in the morning, but generally reliable climbs by the afternoon.
• Machined and poured new parts for the back wings.

July 18, 2007

• Carefully installed my two different types of wing components on SB2. Initial testing looks promising, although I wasn't able to put all four on because the back legs have 3 of the 5 features inverted.
• Learned a little bit about the programming involved in SB, but was not very successful in testing. In the end, we decided to remove them and try to incorporate them by next Tuesday.

July 17, 2007

• Pried out last set of wing parts from wax block. I had used duct tape to cover some of the holes in the block, and this duct tape was covered in plastic during the pour. As I didn't want to plane the surface when it had duct tape on it (can't be good for the sharpness of the planing tool), I had to plane down to half a millimeter, snap off the plastic, and pull off the tape. In the future, it would be better to arrange my parts such that I do not have to cover holes and can instead work around them.
• Constructed four of each of two designs (which have two different stiffnesses). Each includes a servo horn embedded in the compliance piece.
• Ordered additional hall effect sensors -- we're down to two.

July 16, 2007

• Poured final batch of wing components and let them cure during the day. Went back in the late afternoon and the hard polymer still seemed flexible so I decided to let them continue curing overnight. I will pry them out tomorrow morning.
• Drilled out holes by hand on both the compliant pieces and the wings. Will have more to drill when my pieces are pried out tomorrow.
• Worked to assemble one each of two of the different designs. Ended up using different servo horns that initially expected, so the horn grooves actually became somewhat detrimental (removing area to which I could glue the smaller servo sizes). However, this also allows for me to invert the orientation of the compliant piece so that the wires connecting to the hall effect sensor will emerge on the wing side (where it will connect), as opposed to the head side.

Week 3

July 14-15, 2007

• Coordinated with Barrett to share the Haas for the weekend to cut our various parts. for the SB dress rehearsal.
• Cut and poured the soft polymer.
• Pried out the cured pieces to prepare block for planing.
• Machined the rest of the hard parts -- turned out that the cutting length of the short tool was actually longer than initially thought, and with some minor tweaks I was able to use the short tool to cut them and decrease the cutting time significantly.

July 13, 2007

• IE-20AH did not cure -- had the consistency of chewing gum. Most likely cause was use of old polymer -- apparently it expires after three months, so it's paramount to check the expiration date. Spent the morning cleaning it out of the hard parts and salvaging what was possible.
• Made some minor tweaks to the parts and redid the Unigraphics in two sets: pieces that needed both hard and soft polymer/shallow pieces, and pieces requiring the use of the long 1/16" tool. Cut the first group late in the afternoon and was able to pour before leaving.
• Left remaining toolpaths (the soft polymer and the longer tool ones) on chewie for use over the weekend.

July 12, 2007

• Planed hard polymer down to wax surface.
• Acquired a 1/32" extended endmill from Western Tools in Santa Clara.
• Learned to change tools in the Haas, and loaded the new 1/32" tool.
• Machined soft polymer wells into hard plastic -- four parts took approximately 2 hours.
• Poured IE-20AH -- takes approximately 12 hours to cure.

July 11, 2007

• Remachined wing compliance components on the Haas.
  • Took significantly longer because I had to use a longer tool for the endpin depths. In the future I may want to do them as a separate file, because the short tool can handle the "standard" limb depth of about 3-4 mm.
  • Experimented with selective part making using Solidworks and UG to speed-up Haas process.
• Learned how to pour hard polymer. Poured my double set of parts -- they will cure overnight.
• Learned to change the oil in the Haas.

July 10, 2007

• Machined first parts on the Haas.
• Had some difficulties because I had to change the tool at the last moment (I had planned on using a 1/32", but it was broken so I had to replace it with the 1/16") and while I thought everything was okay, there were some sections that were uncut. I also mis-labeled the floor, so some parts were not cut deeply enough. I decided to redo the part files in Solid Works/Unigraphics and then to remachine over the parts that I already have. The files are now ready, and I'm just waiting for machine time on the Haas.

July 9, 2007

• Continued work on Solid Works and Unigraphics for SDM of wing prototypes.
• Lessons:
  • Place parts close together in a unit that can be translated across the wax block in Unigraphics -- do not try to make multiple copies of the part in Solid Works, or to translate each part separately.
  • Use the wax block file from Chewie.
  • Write down all mating geometries in logbook to ensure easy duplication. Also, make note of translational geometries from Unigraphics.
  • Double-check the floor -- make sure it's actually the lowest point.

Week 2

July 6, 2007

• Continued brainstorming different ways to incorporate the hall effect sensor into the wing parts that already exist (or with subtle modifications).
• Talked to Barrett to set a timeline for completion of the parts so that they could be incorporated before SB's rehearsal day in about two weeks.
• Completed Solid Works files for parts to allow for force sensor to be incorporated into the wing-angle of SB: CompliantWing.

July 5, 2007

• Lab cleanup day

July 3, 2007

• Redirected towards designing wing compliance: incorporate a hall effect sensor in the wing to determine the angle offset and ultimately the force in that direction.

July 2, 2007

• Continued work on torsional spring idea, including research into the use of torsional springs, specifications, and types (machines versus wire wound). Worked with Solidworks to create a better model.
• Discussed method to measure resolution and set requirements for hysteresis with Barrett -- he suggests use of micrometer table fixture to incrementally move the sensor between the magnets in order to get accurate readings from an oscilloscope. Apparently the sensor has about 0.02 V of resolution, so I'll need to map that to a rotational distance. The ideal would be to minimize the hysteresis to within that range, but at the very least I'll be able to measure the current design and the proposed designs and see if there is any improvement.
• Moved towards making SDM version of the three "easy" design options: reducing face friction, increasing joint cross sectional area, and increasing joint tension. Was stalled when I could not find Sangbae's original arm link part that contains the mounting for the hall effect sensor.
• Ideas for consideration:
  • Are there ways to improve the current set up for the hall effect sensor (currently it has limited range and because of the rotation the flux changes as the servo rotates)?
  • Flexural pivots as a possible soft polymer replacement (http://www.pages.drexel.edu/~garfinkm/Pivot.html)?

Week 1

June 29, 2007

• Continued work on HysteresisReductionInCompliantShoulder by implementing some of the more basic ideas (ones that do not require significant redesign) in Solidworks.
• Discussed joint dimensioning/design and other possibilities for designs with Sangbae.
• Continued brainstorming.

June 28, 2007

• Began work on reducing hysteresis in the compliant shoulder by brainstorming possible causes and ways to mitigate them.
• Found the current Solidworks files for the compliant shoulder.
• Started HysteresisReductionInCompliantShoulder

June 27, 2007

• Learned about Barrett's fabrication techniques for the two-toed foot design.
• Was introduced to the suite of software programs used to specify SB's gait and to translate it into a program that the robot can understand. Also learned about the ongoing issues/tweaks/advancements for SB.
• Spent the afternoon sifting through the software files on Chewie to get a feel for the code, as well as browsing the Twiki for information on the force sensors and controls for SB.

June 26, 2007

• Watched Alan's preparation of Solidworks and Unigraphics files for SDM, and observed the setup of the CNC mill component.
• Brainstormed first generation implementation for Zman -- focusing on how to switch between dry adhesives and spines for climbing.
• Prepared a Solidworks model and Power Point slide for inclusion in a Zman presentation for Draper Labs.

June 25, 2007

• Completed some of the advanced Solidworks tutorials
• Attended SURI program opening lunch
• Worked through the SDM tutorial, including making the wax block part and transferring to Unigraphics. Had some difficulties with the Unigraphics part of the process, but observaed Sal's process as he made a shoulder compliance part, created the program with Unigraphics, and then began to machine it in the RPL.