Team Five/Journal
From Maslab 2006
| Maslab teams |
|---|
| Team 1 · 2 · 3 · 4 · 5 · 6 · 7 · 8 · 9 · 10 · 11 · 12 · 13 · 15 |
| Team Five's Journal · Paper |
Contents
|
Journal Entries for Team 
December 17 2005
The team had a preliminary meeting at Simmons Hall. The team discussed the strengths of the team and brainstormed ideas for attacking the general problem (even though details of the competition were not released). Shreyes informed the team some of the dates he wouldn't be able to make due to other conflicts.
Christmas Vacation
The team looked and read over the tutorials posted.
WEEK ONE
Day 1
January 09, 2006
Lecture 12:00-~1:00 PM LAB ~1:00 PM - 5:30 PM
The course had its first lecture today as an introduction to MASLAB. And then, the team headed to lab where we finally found a corner workspace where we could work. After picking up the tub of material, Vinith and Shreyes started work on soldering the OrcBoard and Lohith and Surasak (Fa) attempted to build the pegbot. After a long time trying to figure out what to do with the pegbot, Lohith and Surasak (Fa) successfully built the wheels and attached the motors. Eventually, both Lohith and Surasak (Fa) ran out of luck and had to wait for other teams to finish using the correct sized hexnut wrench to tighten the wheels to the peg board. Vinith and Shreyes diligently finished the soldering and was free to get the remaining hardware, including OrcPad, Eden CPU interface, and the LCD screen. The team was quickly able to hook up all pieces together and had a fun time trying to boot the OS interface with the OrcBoard. Shreyes finished the days requirements by writing short-and-sweet "Hello World" programs for both the Eden and OrcPad while Vinith, Surasak (Fa), and Lohith carefully screwed all pieces together to maintain balance. Shreyes also setup a repository using Subversion for all of the code, and stuck the HelloWorld program in there for kicks, although don't expect it to show up in the competition. The team then carefully packed their robot for storage.
Day 2
January 10, 2006
Lecture 12:00-~1:30 PM LAB ~1:00 PM - 7:00 PM
We worked on incorporating an IR sensor into our robot today. We had some problems at first, but we managed to make it work around 5.30 pm or so. Now, our robot can move forward until it sees the obstacle within 6 inches; it then stops and reports to the botClient. After that, we discussed our design and tried to connect the camera to the robot. Unfortunately, it's not working yet and now it captured a black image of nothing.
Day 3
January 11, 2006
Lecture 12:00-~1:00 PM LAB ~9:00 AM - 6:30 PM
The team started early work today in anticipation of Checkpoint One. Shreyes and Fa quickly got the camera working (magically). Vinith wrote a MATLAB red-ball finding algorithm and then incorporated it into Java. Eventually, most of the team got to the lecture late but were able to reach the lecture. The team got together to work on red-ball finding. Shreyes and Vinith worked on improving the red-ball finding algorithm while Fa and Lohith wrote and tested on rotational movement of the robot (which turned out to be quiet fun when trying to determing the best speed possible). The team decided to rotate one wheel forward and the other backwards for a sharper rotation. After both algorithms were matched and compiled, the team had some fun debugging red-ball finding. In particular, the team had trouble adjusting the speed of rotation (if too fast, the robot would stop a little after seeing the red ball, and if too slow, the robot was no fun :-/), and visualizing the picture on the BotClient. After that, Shreyes left early while the rest of the team tried to test the robot on the field, with much trouble because (maybe? we never found out why) of poor wireless connection from one room to the playing-field room (Anders suggested this possibility). Soon, the team tried the robot in the 6.001-LAB field with no avail. Tim checked it out and recommended rebooting and highly recommended us to mount our LCD console in a more stable way. After fixing the LCD console (with an awkard screw attachment), Yoyo came by to recommend possible solutions to our (maybe?) weak wireless connection. Fortunately, the robot worked on the playing field and we passed the checkpoint. Fa and Lohith passed it off as a random incidence, that described best by a geometric random variable. Fa and Lohith bailed out for the night while Vinith continued to hack away at the mechanics of the robot (in consideration of Checkpoint 2) late into the night...
Day 4
January 12, 2006
Lecture 12:00-~1:30 PM LAB ~9:00 AM - 6:30 PM
Today's Lecture was on Robot Complexity and Strategy. After lecture, the team diversified: Shreyes and Vinith worked on the days objectives and Checkpoint Two. Shreyes and Vinith wrote a bar-code interpreter and a red ball finder code. Shreyes and Vinith had some trouble with correctly interpreting the bar code but by the end of the day had it almost done. Fa and Lohith worked on the body and chassi of the robot. After a short discussion on mechanical strategy, Fa and Lohith built a ball collector on the bottom of the robot shaped like a box. That took up most of the day's work. At the end of the day, the team mounted all the pads and the computer onto the body.
Day 5
January 13, 2006
LAB ~10:00 AM - 6:30 PM
We checked off the barcord recognition and worked on check point 2 today (wandering around and stop when finding the red ball). For some reasons, the IR sensor didn't work, so we spent a lot of time figuring out what's wrong. But we still couldn't fix it. By the way, we also started on yellow goal recognition code.
Day 6 & 7
January 14-15, 2006
LAB ~3:00 PM - 5:00 PM
The team tried to get checkpoint 2 on Saturday but weren't able to because wires attached to the motor came off at the most inconvenient time possible. Lohith tried to get the FindGoal class working which acts as an intermediate state between deciding to score and depositing the goal. By Sunday, FindGoal class seemed to be doing the job Lohith wanted originally. Fa tried hard to get the robot moving for checkpoint 2 and unfortunately found out from Yuran that our USB port hub was dead, rendering the camera useless. The team decided to do a temporary fix using Fa's USB Hub from home until monday. Fa also created the final state machine class DepositBall and thought ahead for the Design Review coming up this week. Vinith worked on refining the red ball finding and wandering algorithms. Unfortunately, there wasn't much achieved this weekend and we hope to get up to speed starting Monday.
WEEK TWO
Day 1
January 16, 2006
LAB ~10:00 AM - 8:00 PM
Today, the team focused on finishing checkpoint 2. At the start of the day, Lohith and Fa worked on having the robot move around the maze without getting stuck. Of course, there were sharp corners on the field in the 6.001 LAB and the robot got stuck many times. After working out ad hoc solutions for those small problems, Vinith, Lohith, and Fa worked on getting the robot to aim at the red ball and to move forward. Shreyes wrote up a design plan and planned out the mechanics of the robot. Fa ended the day by having checkpoint two done (finally).
Day 2
January 17, 2006
LAB ~10:00 AM - 7:00 PM
The team started work on the mechanical design of our actual robot. We built the circular body and made all the correct holes for wheels to fit in. In addition, the team created the team design review presentation.
Day 3
January 18, 2006
LAB ~11:00 AM - 8:00 PM
The team presented the design review and completed work on the mechanical design. They encountered many problems, including a good place to put the IR sensor as well as drilling in awkward places. Eventually, the team managed to finish the entire robot. Currently, the team is still trying to debug and fix their code, their IR (the most annoying sensor the team has used), etc.
Day 4
January 19, 2006
We continued working on the image processing code:FindBall, FindGoal, ReadCode (barcode) and FindDiscontinuousLine. Also, we took a lot of pictures from the field in 26-100. These pictures will be very useful for testing our codes.
Day 5
January 20, 2006
LECTURE 12:00 PM - 1:00 PM LAB ~11:00 AM - 6:00 PM After a neat presentation by James, the team headed to lab to work on the software. Lohith tried to fix his FindGoal class, which worked fine for only one goal in one image but wasn't perfect for the case where there are two goals in one image. Vinith decided finally to write up a new algorithm that is at half the resolution to find goals quickly and also, in the case that there are two goals in one image, to store goal objects into Vectors. He also added to Fa's work with respect to the discontinuous blue headers on the walls.
Day 6 & 7
January 21-22, 2006
LAB ~12:00 PM - 6:00 PM The team met at Simmons Dorm to discuss where the team's robot was headed, in terms of algorithms, mechanics, and ball-catching mechanisms. At lab, Lohith and Fa worked on adding the new gate mechanism that opens up to allow balls into the trunk and also closes to keep them secure. Although this is a simple idea, the team hopes its simplicity will allow the robot to have more freedom with ball-catching. Fa and Lohith also worked on adding a taller support structure for the eden and OrcBoard (which is now vertically placed on the body of the robot) so as to allow the camera to move away from the edge and still fit snugly under the vertical, raised eden/OrcBoard configuration. Due to time constraints on the lab hours, the team wasn't able to finish completely securing the support structure. Fa also found out that the gate needed to be lower and fastened tightly at the open end so as not to allow balls to slip out.
WEEK THREE
Day 1
January 23, 2006
LAB ~11:00 AM - 9:00 PM
The winter storm delayed arrival but the team was able to meet at different times of the day. Fa and Vinith worked on software, while Lohith worked on tightening the support by adding metal brackets. Fa cleaned up much of his code of FindBall3.java which scans row by row to find the TOP of any red ball for measurement purposes. Vinith worked on the camera feedback mechanism which the team discusses during the design review.
Day 2
January 24, 2006
LAB ~11:00 AM - 9:00 PM
Fa was still working on FindBall3.java with a new algorithm. Lohith and Vinith worked on getting a good feedback mechanism for the robot to find a goal and to head towards it.
Day 3
January 25, 2006
LAB ~11:00 AM - 9:00 PM
Lohith and Vinith had the feedback mechanism finally working and were able to successfully deposit a ball. Lohith attempted to test the servo for the gate and Vinith tried to write a good ball-catching and wandering algorithm. Fa worked on finding discontinuous blue line algorithms.
Day 4
January 26, 2006
LAB ~11:00 AM - 9:00 PM
Today, the team had their first real mock contest. The robot was only able to rotate twice and got stuck on the wall, possibly because of a gap between walls. Lohith and Fa took some pictures and were able to load them up into the repository. Fa continued to work on fixing the gate and managed to allow the gate to move more smoothly. Fa also worked on FindBall. Lohith attempted to come up with a good algorithm for feedback control on turning around a wall. Vinith continued to refine the robot control.
Day 5
January 27, 2006
LAB ~11:00 AM - 6:40 PM
The team worked on FindWall. Fa finished it such that it returned the bottom position of any wall in a picture. His algorithm used a linear relationship between different wall heights at different distances from the robot. Lohith worked on improving the Gate, making it sturdier and stronger, having it tightly fit with the servo and improving the metal gate catcher/holder. He realized that for the gate to properly fall into place into the gate catcher, the servo had to be tightly screwed into the side of the robot to prevent any angular deviations. In addition, Lohith created a simply and temporary gate class which opened and closed and had respective functions. Vinith did a lot of stuff related to navigation.
Day 6 & 7
January 28-29, 2006
LAB ~? - ? The team met to do some navigation stuff... during the working phase, aliens came and abducted the team and experimented on them and then returned them on Monday morning. Indeed, 42 is the answer to THE question.....
WEEK FOUR
Day 1
January 30, 2006
LAB ~10:00 AM - 3:00 AM
The team had mock contest 2 today. Our team scored 0 points, both during the actual run and the extra run we tried. But hey, this placed us in the top 5. The team tried to make the robot more agile in turning and getting out of corners. e − | t |
Day 2
January 31, 2006
LAB ~10:00 AM - 10:30 PM
The team built a simple gate system with two coupled servos. There are two blocks of wood that rotate with the servo in a synchronized manner to allow balls to go in. When the gate is closed, there is a small gap between the pieces of wood (small enough to not allow any red balls to go through). The team then continued to score balls into the goal using the new gate mechanism. After the gate mechanism was calibrated, the team continued to try to catch red balls using the robot. Unfortunately, the bot always deviated in one direction for no apparent reason. Much of today was dealt with clearing up this tragic issue.
Day 3
February 1, 2006
LAB ~10:00 AM - ???? NEVER ENDING STORRRRRRRRRRRRRYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYY
Today, the team realized that the robot would be better off with haveing the camera adjustable but tilted downwards so it could see more of the field right ahead of the robot. The team decided to build a sturdy wooden bridge that held the camera in place but allowed it to be tilted forward. This greatly improved the accuracy of the robot but other issues regarding color thresholds and calibration came up that had to be resolved.
So technically now it's no longer February 1st, but we really count days be when we sleep, and we're not sleeping yet. Hopefully we will before the competition. We learned many lessons today. First, we wasted a LOT of time in the past two weeks on image processing, a lot of which we're actually not using now (or have changed). If we're only using the camera for movement, then it's kind of important that its mounted well. Our fps was extremely low, which explains why our feedback during the driving process sucks. It'll get better, we have about 27 hours. The lights on the coke machine outside don't work. I did find $.75 though in the coin return though, email the team to claim.
5 am now, of course you would know that though if you were checking the history on this page. So major bug fixed in ImageProcessor, major annoyance fixed in driving straight, our gate sucks, our feedback rules, we can catch balls reliably, and our robot is happy. Now on to implement the code we wrote 12 hours ago, i just hope we remember what it was actually about.
Day 4
February 2, 2006
LAB Never Slept - 5:00:00 PM
Judgement Day: The tEam corrected their ball finding algorithM to ignore bAlls near or Inside the goaL. The team continued to test this with every singLe picture they had taKen from past mock contests. The team then contInued to work oN theIr @ FindEdge algorithM so that the robot would go to edge and back up when IT crashed. EDU (Look at the clues ;-))
Day 5
February 3, 2006
Judgement Day II: All audience members interested in our team's journal should try to solve this simple puzzle and email the team maslab-2006-team-5 AT mit DOT edu : Find the odd one out and explain why:
FRAP ITAR PERL ROMB JAPT SPAD
