Team Two/Journal
From Maslab 2013
Day 1 (January 7)
- Lecture 1 gave us an introduction to MASLab.
- Lecture 2 was an introduction to programming our robos
- The best part was the lab:
- 1. We assembled the PegBot.
- 2. Coded the robot using Arduino to move at various speeds and directions.
- 3. Didn't face any problems yet, except that the left wheel is a bit too close to the body of the bot. Didn't affect the movement much at high velocities.
Hungry for more!
Day 2 (January 8)
- The lectures were on sensors and algorithms.
- Our MechE member received Edgerton Shop training today.
- We fixed the issue with the wheel scraping the pegboard.
- We discussed the design of the end effectors.
- We also discussed the general scoring strategy.
Day 3 (January 9)
- Another team member got shop trained today. Added helping hand!
- Today was planning day. The following plans were laid out:
- 1. We planned our calendar. We have decided to have the robot mechanically ready by week 3. In the mean time, our coders will decode the mysteries of image processing and mapping algorithms. Post week 3 will mostly be debugging time.
- 2. We finalized our team strategy: Grab, collect, store and go for the pyramid. Aim for the wall only if we start running out of time.
- 3. Our MechE designed the robot on SolidWorks and our bot looks mean.
- 4. We designed the state machine; foolproof as of now!
Day 4 (January 10)
- Lecture was on mechanical design.
- We finished half of the mechanical design details.
- We transitioned from Arduino to Python.
- We started using openCV to handle vision. We were able to filter out colors in HSV.
Day 5 (January 11)
- Craziest day till now.
- We burnt our Arduino board and our Motor Controller.
- This made us rebuild the bot from the scratch.
- The outcome was a bot with clean circuitry and added mechanical stability.
- The original caster wheel was acting crazy so we replaced it with a ball caster.
- Got the bot to follow a red ball.
Day 6 (January 12)
- We finally figured out the nuances of OpenCV.
- Calibrating the camera seems so easy now.
- Bot now follows any specified color and under any lighting condition.
Day 7 (January 14)
- Mock competition 1
- Ok so the bot follows a ball now what?
- We finalized our bot's explore state --> wall following
- Used two IR sensors and a bump sensor to maintain the bot in a state of exploration.
- Bot starts with a button start and stops after 3 minutes.
- We were happy with how the bot performed.
- The main take away message for us was: sensors are really important; use them wisely to keep the bot from getting stuck.
Day 8 (January 15)
- We are trying to enhance our image processing techniques!
- A lot of noise in the image causes our robot to lose sight of the ball
- Also, our explore state is not well defined
- Plans are to use wall following as the explore state
Day 9 (January 16)
- Another major issue in image processing seems to be deciding which ball to go to in a situation where multiple balls are viewed.
- Currently, the image detects the centre of mass between two balls when they exist in the image and then tends to move towards the centre of both balls.
- We implemented a PID controller, which has stabilized the movements of the bot greatly.
Day 10 (January 17)
- We completed all laser cutting necessary for making a drivable robot for Mock 2.
Day 11 (January 18)
- We ran into a problem with the timing belts. McMaster recommends sizing timing belts 10% smaller but the belts do not stretch even 1%
- Everything was assembled but we were unable to drive the robot
- After Mock competition was over, we drills new holes on the side walls to reduce the shafts' center-to-center distance
Day 12 (January 19)
- We fabricated the sheet metal scoop
- There was interference issue with the rubber band roller axle so we set the scoop back by 0.5"
- We mounted the drive motors to their new locations and installed bearings to the wheel axles.
Day 13 (January 21)
- We replaced the rubber band roller's cardboard spacer with a metal one and permanently fixed the roller to the axle
- Rubber band roller motor was mounted and its transmission set up
- We re-wired everything to clear the clutter
- We tested the rubber band roller--success! The robot was able to pick up 4 balls abreast as designed.
Day 14 (January 22)
- We laser cut the pieces for the scoring mechanism
- Finalized the code for Mock 3
Day 15 (January 23)
- Mock competition 3
- Run 1: 1 ball collected
- Run 2: 4 balls collected
- Happy with how things worked today.
- Image processing still seems very slow. Down-sampling the image received may be a a probable fix.
- Need more sensors.
- Code needs to run more threads.
