Team Four/Final Paper
Contents |
Introduction
This years MASLAB competition was different from the past years. Instead of creating robots to collect balls and deposit them at different spots on the map the competition this year was to stack 2"x2" cubes at different locations on the playing field. There were three different scoring regions and various different bonuses for stacking cubes in specific areas. After discussing a strategy with the team we settles on a fairly simply strategy. We decided to stack towers both in the field and in our hombres. This meant that we would be able to stack both colors of blocks and also get the multiplier for scoring two different ways. We also decided to pick up blocks and store them on-board and later stack them on different parts of the map. After creating a strategy for the competition we needed to come up with a mechanical design.
--Important
When we were designing our strategy (and our robot) we thought that the stacks had to be left self-standing while the robot was able to drive away. Because of these constraints our robot had a very unique design that allowed us to stack blocks in two separated containers and then drop the towers and drive away. If the rules stay the same next year we would recommend creating non-moving towers which also allows the pick-up mechanism to be simplified.
Mechanical
The first task for the mechanical team was to brainstorm and model various different ideas for a robot. We began our brainstorming session by first creating a strategy for the competition. After coming up with our strategy (which you can read about above) we started working on a majority of the CAD design. Our goal was to have the robot completely designed within the first two weeks of class. This goal, while ambitious, is definitely do-able and important. If we had continued pushing off our mechanical design the software team would never have had the time they needed to complete the software side of the project.
It was also important that the mechanical team was comfortable in the machine shop using all of the tools. This included band-saws, sanders, drills, wrenches, and a laser cutter. Without feeling comfortable around these tools we never would have been able to quickly prototype different mechanical designs. The ability for us to prototype quickly was extremely important and is what allowed us to finish a majority of the mechanical design in the first two weeks of MASLAB. It is also important that the mechanical team is comfortable quickly adding parts/fixing parts on the robot because it is bound to break many times.
Electrical
The most important electrical goal to keep in mind is organization. From the beginning we kept our wires labeled and on a specific color pattern which allowed us to easily debug problems with the robot. We would recommend that teams next year focus on keeping the electrical aspects of their robot neat and organized. By labeling all of the wires, keeping a specific color pattern, and making sure to wind wires your electrical work will be much easier.
Other teams also had problems burning our their Edisons. We would recommend attaching all the 5v power needed for the servos and sensors to an external DC-DC converter so that if there are any shorts the converter blows and not the Edison.
Also make sure to charge your battery daily and DO NOT forget to plug in the LiPo smart connector. If not, your robot may easily drain the battery which can be dangerous if they are again plugged into charge.
Software
We used a combination of computer vision, environmental sensors, and motors to control our robot. The computer vision was used for the most important aspect of our project -- sensing cubes. Our strategy was to drive around (wall follow) using analog short-ir sensors. We took pictures every so often and analyzed them to see if there were any cubes in-front of the robot. After find the cubes the robot would drive up to the blocks, knock them over, and then drive over single blocks and then picking them up.
One of the major flaws in the software was the color detection of blocks within stacks. Although the color detection work beautifully the camera sometimes missed the color of the block it ran over and thought that it was a different color than it actually is. If we could do it again we would have hooked up a color sensor on the interior of our robot to recognize what color is being picked up and therefore put it in the wrong stack on the robot.
