Team Ten/Journal

From Maslab 2006

Maslab teams
Team 1 · 2 · 3 · 4 · 5 · 6 · 7 · 8 · 9 · 10 · 11 · 12 · 13 · 15
Team Ten's Journal · Paper

- go big or go home - 
February 1, 2006
1. I can't really remember much of this day. I know alot happened, and the robot is
   sort of coming together. Team is not sleeping much

- if only it was like this a week or two ago -
january 31, 2006
1. scanning and getting balls is in an initially acceptable state. 
2. wall following is really hard. we decided we didn't like the placement of our bump
   sensors, so we took them off and decided we'd work on them the next time the shop
   was open. So, we tried to wall follow with just our IR's, which proved very 
   very hard. Robot handled corners poorly. maybe our IR's placement is a poor choice...
3. check point 3 completed. staff present "would have liked to see better wandering" 
   but we can get balls and take them to goal. 
4. really, there is not enough time. Proceeding as though we will complete most tasks 
   need to. 

- 3 days? - 
january 30, 2006
1. Possibly to little to late, code is finally cranking out. At roughly 9pm we scrapped
   most of our written-before-testing bloated multi-threading code and decided to just
   "make something work sort of." Vision stayed, and has continued to be cleaned. we're
   updating image processing @ ~130ms/update. No more multithreading though, at least 
   not for a while. 
2. robot still not perfect! exasperation!!!

- dug in a deep hole - 
january 29, 2006
1. We are visual servoing to objects with regularity. Calculating angles from balls
   is still a bit buggy, so I think we'll be scrapping calculated aims to objects 
   and instead opt for a constant small angle turn before rescans. Its less cool
   but we are clearly playing catch up. 
2. Blew a short range IR and seemingly port 0 on our robot I/O . woops. 
3. all following strategy is turning out to be kind of scary. Especially since our
   entire robot functionality depends on it. Advice time: if your going to make your 
   robot depend on a certtain functionality, make sure you can do it already or have 
   done it before in a different situation. This is kind of scary, but we may overcome. 
4. debuggin visual servo took to long so we didn't actually implement wall following code.

- actually testing - 
january 28, 2006
1. Bar Code checked off. (what, like 2 weeks late?)
2. Day one of testing is mostly depressing. Not actually depressing, but it hurts 
   to see that robot is not as mechanically or softwarically coherent as we might 
   have expected. New opportunities.
3. need bump sensor and ir sensor additions, 

- test 1,2 -
january 27, 2006
1. Robot is complete enough to start testing. Still didn't actually do any testing,
   but it's coming. 
2. There's are some comedic jamming scenarios for the ballLift function (that raises
   the balls to get into the scoring queue). Comedic when there is still 5 daysd until 
   impounding. Hopefully these situations can be solved mechanically, because we're 
   low on time and sensor points to implement any smarts in our ball scoring mechanism. 
  
- oops on the mock contest - 
january 26, 2006
1. had a sort of functional robot, but missed our chance at running in the mock
   contest. Sorry, Staff! Looks like few teams were well prepared for the contest, but
   at least they had team members in 26-100 at the correct time. Again, our apologies.  
2. Version2.0 is functional!:
     - it's cute. still need to work out motor mount for the lifting mechanism. There 
       is a mount in place right now but it really impedes ball travel in the ball 
       coral. If this mount doesn't figure itself out, the ball coral is next to 
       useless. 
     - Encoders still not functional.
     - placement of IR's for smart wall following now necassary.
3. explanation on current functioning and planned threads could be coming soon. 

- version 2.0 -
january 25, 2006
1. robot redesign and build is well on it's way: 
     - passive ball apprehension has an 11.5" window in front of robot. This allows
       for a significant amount of inexactness when apprehending balls. 
     - high-torque motor used to lift balls up into the exit queue. Rough estimates
       say that it will take ~20 seconds for two balls to be lifted from the ball 
       coral into the queue. Once in the queue, balls have potential energy and need
       only be let out with the flick of a wrist (servo). The queue can hold 4 balls. 
       There may be significant issues with it, engineerically. 
2. vision class quite functional:
     - imageprocess ip = new imageprocess (null) ---> ip.process returns a list of
       locations of all visible balls, goals, barcodes & barcode code. There is a bug
       with false positives of barcode readouts (usually resulting in "B B B B B") and 
       barcode-superball interaction, that we probably will just live with. 
3. thoughts on calibrating vision to travel to objects in arcs:
     - would require us to having working encoders (gasp!) 
     - smoother robot functionality, "oohs" and "aaaahhhs" from the crowd, arguably
       more robust apprehension of targets.   
     - need a final robot design, so the camera can be statically mounted once and for all. 

- -
january 24, 2006
1. seriosuly considering robot re-make. ugh. 
2. need to be testing. we're close, but not there. 

- cleaner robot, almost functional - 
january 23, 2006
1. basic strategy rethought: pursuing a goal scoring device that approaches goals
   from the side, instead of the front. This integrates scoring and wall following
   behaviors, which we think will save significant amounts of time on every goal-score.
   When knocking out wandering strategies, we really like wall following, as it
   inherently encourages travelling to knew destinations, which is important if we
   are two explore and find as many goals as possible. Though, no plan is fool-
   proof, or without it's weak points. 
2. Camera mount temporarily permanent. Hard fastened to chassis, with a tilt up/down
   capability. Allows for some fine tinkering w/o requiring re-calibration on the 
   software side. 
3. organization laid out in code for basic functionality. Very close.
4. Hardware goals for near future: finish design of scoring device, possibly find
   some faster drive motors that effortlessly fit into orc motor interface. 
5. software goals for near future: basic wander -> ball find -> ball apprehend -> 
   wander -> goal or ball find -> goal score or ball apprehend. Also, read bar codes. 
6. Even after a re-flash of our orc board, team10 is still getting crazy encoder read
   outs. This is worrisome. 

- weekend - 
january 21, 22, 2006
1. team10 seems to lack the killer instinct enough to robot-binge on weekends too.
2. hopefully, our relaxed, positive states of mind play to our advantadge. 

- coding, mounting - 
january 20, 2006
1. somewhat precise turning behavior via gyro sensing attained. It's the inertia
   after stopping that is variant here. 
2. encoders provide pretty awkward, discontinuous, data. It's a bit frustrating. 
3. chassis seems to attain balls with ease, ball scoring mechanism still on
   drawing board, though worse case scenario we bull-doze with the front gate. 
4. weekend goal: have basic functioning robot (ie. achieve checkpoint 3) 

- it's code time - 
january 19, 2006
1. sticky cork-like material added to wheels. Traction, hopefully.
2. Encoders re-mounted on motor brackets. Less brakey-brakey, hopefully.
3. There seems to be a complete, albeit not finished, robot in our hands now.
    The robot has mechanical wits enough to drive over balls and contain them, 
    the camera is mounted in a seemingly close-to-final destination. The only
    moving parts we have left to add are some bump sensors. 
4. Scoring mechanism is still very naive in that it has no idea of if it got a
    ball it attempted to get or how many balls it has. 6.270 kindly donated some 
    matched IR emittors and sensors that we'll be using once we're sure our 
    mechanism is final.
5. Pending a change of heart, the basic mechanical structure is complete and it
    is time to move into intelligent code design.

- regulating - 
january 18. 2006

1. Gave our "design review" presentation, replete with childrens drawing of our robot. 
2. Encoders mounted, highres encoder pattern mounted on wheels. 
3. New Motor Mounts made, and trimmed to size. 
4. Computer case modified for increased cooling, so our Eden won't overheat and stop working.
5. New camera case created for easier camera mounting. 
6. etc.... Robot is starting to look solid. Hopefully our simplistic design will work to our advantage. 

- a little behind - 
january 17. 2006
;not working very hard over the weekend, team10 plays catchup;
1. Encoders about ready to go (been assembled).
2. Servo motor control explored; mechanical gate designed:
      - Servo class as described by orcboard api is fairly self explanitory.
      - Gate looks to be the only moving part on our robot, other than
        drive wheels. 
3. Chassis, with tracks, assembled:
      - Inverting motor mounts from their "standard" configuration gives us
        enough ground clearence to drive over balls with ease. Balls fall into
        specific divisions under chassis that arrange balls into a line, 
        allowing for more simplified mechanism to score balls in controlled
        numbers.
4. Blue line filtering achieved.
5. Code begins recognizing "maslab-objects" based on shape, color, 
   and orientation.

- it's beginning to look alot like robot - 
january 16, 2006
1. New Computer Case Assembly:
      - Designed and assembled Eden case out of clear plastic. It looks clean, 
        and ought not to jingle and jangle like the old case did. Also clears up
        some of that "my computer is mounted on a conductive surface" anxiety.
2. Ball Picker-Upper Apparatus Assembly:
      - Yet to be thoroughly tested, expect to be able to naively gain control 
        of balls in a manner condusive to scoring points. 
      - Expect to be able to know and control "number of balls in robot's lair," 
        on ball entrance and exit. 
      - May have to stretch our sensor points to get any decent SuperBall 
        manipulation.
3. Complete Basic Robot Design Complete (in dream form):
      - pictures, diagrams to come...

- stoopid robot -
january 13, 2006
1. The robot is very very dumb. Rudimentary wandering and ball-attacking 
   code got through checkpoint2, but the code and robot behavior is not 
   robust.
2. Weekend goals: fresh chassis, encoding incorporated into motorControl class,
   Possibly a more working outline of desired behavior, actions, and reactions.

- ... -
january 12, 2006
1. Have code that reads a barcode off of pictures on a hardrive. need to reroute 
   camera data from the eden into this program.
2. Writing basic mobility class.

- regular progress -
january 11, 2006
1. Spent close to two hours on hardware issues:
       - The eden was only sometimes turning on, though the orcpad always did.
         solution: Some of the solderjoints on the eden were loose =(
       - Blew another battery-fuse. We're averaging 1 a day. 
2. Saw Red Ball:
       - Completed checkpoint1: robot turns around and round until it identifies
         enough red pixels to be something useful (ball). Essentially, we took
         the imagetutorial code and added a loop to see out of a camera through
         the eden. 
3. Beginning to identify and code overarching robot architecture. A little bit. 

- technical difficulties -
january 10, 2006

1. Experienced alot of frustrating and time consuming technical mishaps:
       - Our battery blew a fuse for apparently no reason early in the day
       - Laiter, on the mock playing field, we knocked down a wall (instead
         of stopping in front of it) which proceeded to fall on top of the
         bot and shorted our board and blew another fuse. (we think)
       - Had mysterious trouble getting on the wireless network in the morning
       - Lost a hardrive (for no apparent reason), losing our check-point-1 
         code and the shell script we had for remotely controlling our robot
         via ssh/keyboard. Lost hardrive in the process of backing it up...=(
2. Did "accomplish" some things despite the barrage of setbacks:
       - BotClient is functional on our computers
       - shortRange IR sensor is reporting similar values repeatedly; although
         the IR seems to treat "infinty" like it treats close range, and it 
         does not function very well on shiny floors.

- hello, world! -
january 9, 2006

1. OrcBoard, OrcPad, Eden, PegBot recieved / assembled.
       Team10 believes that the computer case and motor mounts that 
       we recieved could be improved on significantly.
2. helloworld code run.
3. a bit of java head scratching.
4. remote-control robot h4x....
       wireless campus + robot with good low level interface and 
       wireless card + ssh = robot partay.
5. brainstorm..