Today we had a conference call with one of our alumni, Fletcher Porter. Fletcher previously took a class on PID controllers and was happy to spread his knowledge on them. We contacted Fletcher in hopes that he would help us solve our overshooting issue on the arm. Our main issue with the arm is that it keeps traveling past its target position. This issue causes much grief during competition because it makes our driver miscalculate and sometimes even lose the minerals in the sorter.
Fletcher started off by explaining what PID controllers are. PID stands for Proportional, Integral, Derivative. These are tools the controller uses to compile past (using the integral), present (using the current position proportional to the target), and future (using linear approximation) data to move the arm as precisely as we can. Using a PID controller will negate this issue we have with the arm and improve our performance.
We continued our process dismantling the old robot and started work on the new robot. One change is having a golden stain to make the robot more aesthetically pleasing and to highlight our design choice of wood. A physical change was altering the center of gravity in order to negate the chance of flipping while crossing over the crater. We also changed the shape of the bottom of our side panels to avoid getting stuck on the crater. We have made a lot of progress, but there is still a ways to go with our new robot. In other news, we have started planning for a skirmish with other local teams and revised and redesigned our notebook. In addition, we discussed possible rewards and our presentations for them.
Today we planned to make our robot less cheap by adding some new features. As it was, we started to dismantle the robot and got ready to replace it with the new robot. The new robot will feature stained panels as well as some other new features planned in CAD. This is because we wanted our robot have a improved look to make it look like a moving piece of furniture. After our visit to iRobot, we are planning to use their techniques to further improve our robot. We also started to look at our reward plan.
Today we did some preparation for meet 3 such as tuning the robot’s mechanics including it’s program. With the new parts attached on our robot, this will make harvesting minerals over the crater a lot smoother. We started by cutting off both bottom parts of the harvester including an inch of the skirt. This makes climbing and harvesting over the crater more smoother. Then included a guider next to the harvester’s deposit hole with a senor placed inside the dumper. When the robot harvests 3 or more silver or gold minerals, the senor will active which causes the dumper to get rid the extra mineral and continues its harvesting and dumping process. Finally, we attached a sonar sensor on one of the side panels for more consistency on the autonomous.
Today we collected the data about how many times a gold or silver mineral would fall out of the dumper while gathering. We did 3 categories: one with the robot sitting still, with the robot driving forwards, and one with the robot driving backward. We did 10 tests per silver and gold combination (gold/gold, gold/silver, silver/silver, silver/gold).
We also tried to gather data on our the overshooting issue with the arm by running trial runs and seeing how many times the arm would overshoot. This is what we found.
After 4 trial runs we found that at 1.0, .9, and .8 motor speed, the arm would consistently overshoot 1-2 times out of 10. This is a serious issue that will cause our driver to lose precious TeleOp time due to wasted minerals that would miss and result in no points.
Today was our second design review. We invited back JPL engineer Mr. Feldmen and Mr. Zarcara from Caltech to judge our robot and our implementation of the designs we spoke about at our last review.
We explained our use of the coin sorter we had shown them at our previous review, and how we were never able to build a linkage that could complete the task of scoring in the lander. Once we finished catching them up on our design, we began to explain the problems we were experiencing and what we could do to solve them.
The first thing we wanted to address was the issue we had with the harvester and how it would sometimes make the minerals fall right out of the robot rather than depositing them into the sorter. We explained that we had suspicions that the sorter could be at too little of an angle and was simply causing the mineral to hit the first mineral and bounce out. After debating about it, we decided to test the robot and observe what would happen. We found that several times, the mineral would shoot out of the harvester very quickly as a result of too much power by the zip ties. We came to the conclusion that the simplest solution would be to simply make the top layer of zip ties shorter, reducing the amount of surface area hitting the mineral on the way out.
Minerals stuck under robot:
Another problem we brought up was the issue of getting minerals stuck under the robot. We showed them a demonstration of this and the scenario that causes this problem. After observing the problem, we found that there was no easy solution to the problem, just ways as to how we could prevent it from happening. We found that by going over the crater very slowly in the beginning of a match, we would be able to slowly descent upon the minerals and push them away from the edge of the crater. This helps us because the edge of the crater is the number 1 spot where we would get minerals stuck under the robot.
Trouble going over crater:
Another issue we had encountered with the current version of our robot was an occasional difficulty clearing the crater. After driving and observing the issue, we found that the problem was a result of bad alignment with the crater. In other words, whenever we would try to go over the crater at an angle, the robot would have trouble going over it. This was thanks to the front of the robot that had particularly low clearance. Mr. Feldmen and Mr. Zarcara both agreed that the easiest solution to this would be to simply cut the part of the robot that was too low. After doing so, we found that we had much more mobility going over the crater and were able to clear it more efficiently. We decided to update our CAD model immediately so our next cut would have the adjusted panels.
Ultimately, we found the solutions to our problems and also received constructive criticism from professionals in the engineering field.
After having our second table top review, we continued to identify and address the several mechanical issues that were slowing down our TeleOp cycle time. The two mechanisms that we found were having the most issues were the harvester and the arm that raises our sorter.
Today we found that the robot was spilling the second harvested mineral over the first, causing us to only carry one at a time. At first, we believed that our sorter was not at the proper angle to allow both minerals to sit inside it. After a few tweaks to our robot in which cut off a few pieces of the Plexiglas that was blocking the entrance of the minerals, we found that it was, in fact, the harvester that was causing the issues. After close review, we found that the third layer of our harvester was too stiff and exerting too much force, reducing the accuracy and predictability of where the second mineral would land. We solved this by simply cutting the top layer of zip ties to a shorter size.
Another issue we came across was that our scoring arm was overshooting, causing our driver to severely miscalculate and occasionally crash into the lander with the arm, spilling the minerals within the sorter. We were not able to find the origin of the problem but expect to address it in our next meeting.
We were trying to use a force sensitive resistor (FSR) to detect if we have two minerals in our scoring mechanism. If there where two minerals in the scorer it would turn on lights so the drivers can see if they were holding two minerals easier.
We fond that the gold mineral was too uniform to be detected by the FSR if it was placed flat on the floor of Scorer. Then we tried putting the FSR on the left and right side wall which had the same problem. Then we put the FSR on the back wall which could detect the gold mineral, but could not tell the differences between the two minerals types. (This is shown in the graph below) Then finally we put the FSR back to its original position but used a cut zip tie to prop up one side this allowed the gold mineral to deform The FSR.
Decision: We decided to use the final meted of using a cut zip tie to prop up the FSR
Today we came in and assembled the new panels for the robot. After several hours, we finally had the new robot fully assembled. We wired up the robot but did not have enough time to program it and get it to work. We came across a problem that consisted of the motor blocking the arm from being able to raise thanks to very little room between the axle with the arm and the axle being run by the motor. Fortunately, we solved the arm problem by flipping it onto the other side of the panel. This design of the robot and will have changes for improvements in future meets.