Team Eleven Journal

.: 28-JAN-2004 Competition day.

Today was woody's day of glory, and to be honest - I don't think we could have asked much more of him.

He talked - he picked up balls - he even told jokes (pick up lines), while roaming the field). Definitely the most popular robot.

Sadly, we lost by 4 feet - time ran out and woody couldn't find the goal in time. In hindsight, a short term memory possibly making use of dead reckoning - might have helped woody score goals, at the end of the round he was holding 7 balls (more than enough to win), but couldn't find a goal.

Woody and the team win the Engineering award, which is almost better than winning the competition. The staff has faith in woody, his brain (program), and his proud set of 3 parents.

Goodbye Maslab, and hello sleeping in our own beds.

.: 27-JAN-2004 Time keeps on slipping...into the future...

• another night of no sleep, and a dismal morning
• 8:30 am, IR sensor fails and takes a while to debug. Replacement only available when TA's come around
• 9am - hard drive fails and it takes waaaay too long to fix. thank god they had replacements ready to be imaged.
• as of 12 noon, our bot didn't wander correctly, collision avoidance didn't work well, and our ball pickup is still highly problematic. Did I mention that we didn't have docking code?
• emergency meeting at 1 accompanied by dunkin donuts refueling
• decided that dheera and jmw might have been correct with the idea that the software system needed to provide direct access to the IR sensors
• 2:30 roles around and collision detection is working ( agile programming methods make for an easy change to peel away abstraction from IR Sensors, while retaining a quality software system)
• 3:00 - same wall follow code is modified to work perfectly with new IR Data
• 4:00 - we now have a working, wandering bot
• 4:30 - docking code rev .1 is out and being tested
• 5:00 - impounding starts, and we continue trials for as long as possible - adjusting parameters in the code for best trial scores
• 6:00 - after 2 days straight with NO sleep, we turn in a fully functional bot that looks like it stands a damn good chance of doing well
• 8:00 - Viewing of the OC at simmons, followed by a prompt passing out of all team members (dheera did not partake in the OC goodness).

.: 26-JAN-2004

• Start of the big push till impounding

things we need to finish: 1. get belt drive working flawlessly 2. get software system completed and tested (mapping may have to be kept out due to time contraints) 3. make sure our random walk covers entire field (we still need a reliable wall follow - someone code one, it's not hard!)

.: 25-JAN-2004

• Working on getting new code working with old code, and having a fully functional robot
• stayed up all night working on bot, so most of today was spent running trials at the mock contest
• after the mock contest, team members retired to their respective beds ( felt damn good to sleep in our own beds)

.: 24-JAN-2004

• James Mclurkin rules, although his lecture causes less time to be available in lab - damn cool presentation
• stayed until lab close to put final touches on chassis before tomorrow's mock contest
• one long night until tomorrow (no sleep until brooklyn)
• settled on using a belt drive even though it's very sensitive to tuning

.: 23-JAN-2004

• ball pickup still needs to be completed - but main chassis is finished.
• smaller size is causing new problems (for the ball pickup - a little harder to pick up balls)
• numerous other methods are being brainstormed and prototyped -

which do not include the problematic belt drive - (wire elevator, rubber band belt ...etc)

.: 22-JAN-2004

• So it's saturday...and We decided to come to lab anyway.
• jmw and dheera are waiting on the chassis to be finished before more code is written, so that's a priority
• Adamb continues to write code
• chassis still takes time

.: 21-JAN-2004

• began trying to understand adamb's code structure for more advanced contest code (involving features identified by the camera, behaviors that may suggest actions. actions are selected by priorities given to the behaviors such as ball pickup, wandering, goal dropping, etc.).
• jmw continued hacking on new chassis - things are going slower than hoped

.: 20-JAN-2004

• bugfixes to image algorithms (array index checking)
• mock contest 1: hacked code failed due to change in the order of coords in createFeature
• image code seemed mostly okay on 26-100 field. had to widen thresholds for reliability
• decided for sure that old chassis sucks - all or nothing on new chassis

.: 19-JAN-2004

• wrote simple hacked wandering code
• moved work to 6.001 lab for less crowd
• began plans for new chassis design (smaller, lighter, easier to manuver)
• cleared checkpoint 2, able to find balls by wandering

.: 18-JAN-2004

• bugfixes to image algorithms (array index checking)
• assembled all mechanics,

.: 17-JAN-2004

• more work to make ball pickup sturdier so it doesn't kick balls
• wrote barcode-finding/reading code
• calibrated goal-finding code
• rewrote ball-finding code to take advantage of subsampling
• set up laptop to be able to use camera directly to speed up development while mechanics are being worked on
• wrote speech server on laptop so robot can transmit text and text-to-speech can be heard for debugging purposes

.: 14-JAN-2004

• began writing checkpoint 2 code
• wrote goal-finding code
• fixed ball pickup thing to chassy

.: 13-JAN-2004

• adjusted ball pickup

.: 12-JAN-2004

• machined dowels out of aluminum. works better, takes tension but is too hard to turn with rubber bands. belts are better at picking up balls also, but need to make a better belt
• built trapdoor and top tray for ball collection and release
• more chassis work
• discussed software structure

.: 11-JAN-2004

• obtained long rubber bands from staples... figured out that 30 of them puts too much tension on a wooden dowel and will have to re-machine them out of aluminum instead
• using a belt it succesfully picked up balls but the belt slips... rubber bands may be better and will have to try tomorrow
• took apart pegbot and fixed computer and motors to new design
• made motor mount for ball pickup (but it doesn't fit in the tub now so need to put the motor on the inside and use gears
• joystick control program to make it easier to control robot manually from a distance

.: 10-JAN-2004

• began building ball pickup mechanism, lathed wooden dowels for rubber band pickup design
• began building chassis and built framework for ball pickup

.: 07-JAN-2004

• finished checkpoint 1
• started structuring drive class
• discussed vision algorithms
• began building ball pickup mechanism

.: 06-JAN-2004

• wrote algorithm to find closest ball and keeptracking it
• more svn crap
• discussed drive mechansims and mapping strategies
• added feature to enable robot to cook food for us

.: 05-JAN-2004

• finalized software engineering plans
• wrote build scripts for building/uploading jar
• image capture working - sorted out how to get pixel data
• started working on ImageInfo class

.: 04-JAN-2004

• made ir sensor cable
• makeshift pegbot assembled
• working ir bot (detects <.4m, goes backwards a bit, turns in a random direction and keeps going)

.: 03-JAN-2004

• hello world