Overview: In this first week of build season, we all got together for the game release video, read the rules, built new relationships between our teammates, and started prototyping mechanisms based on the way we want to play the game.
Kickoff was a great start to the new season! We all met up in the morning to go over some housekeeping items before watching the game reveal video. After we broke for lunch, we reconvened to read the game manual to get a full and complete understanding of the rules and how to play the game. Throughout the day, we would read for about an hour and break into team building groups. One leader would show the rest of the group how to do the activity then guide discussion. The activities used on kickoff were like the activities used during the Leadership Summit, including human knot and the eye contact activities. New activities included a blindfold activity where people were strapped together and had to navigate through an obstacle course, fitting into a 2x2 and 1x1 taped off box, and moving together in a line while holding foam balls between ourselves. At the end of the day, we met to discuss what we learned from the teambuilding activities and what we can do better for next year's teambuilding. We learned that if we want to go far, we go together, that communication is key to getting a task done, that we all need to contribute and listen to each other's ideas, and we built new connections with our teammates.
On the start of Sunday, we talked about potential strategies for the game season. The overall consensus was making sure we are always getting points during the game - having one robot always shoot balls into the goals, have a robot do defense/bring balls back to the first robot, and another robot shooting balls and managing the spinner when the position/color needs to be changed. During the end game, we would all prepare to climb while the first robot continues to shoot balls, then go to climb at the very end. After strategy discussions, we talked about the prototyping process and broke into groups based on subsystems - spinner, intake, shooter, climber, hopper, and drivetrain.
Spinner: We went through a bunch of different iterations of a belt mechanism to spin the wheel of fortune. We tried using a wheel, we wanted to test timing belts, so we used belts and sprockets from 2018's claw and made a simple loop using shafts, the belts and sprockets, and HDPE that we measured and cut a bearing hole into. After testing that, we refined the positioning of the belts, and swapped the two thinner ones with a single thicker one. We want to spend a little bit of time making a final prototype, since in our current one, the belt is too loose. After, we will start to CAD a more complete design, as well as think about where and how this could be powered and positioned on the robot.
We consulted Brian Sherman about the making and use of these belts and he got me a connection with an alum of 341 Daisy, Tom. We gave him a call and he got me a hand sketch of a basic jig to make the custom belts. The process is simple, but time consuming, heating the belt and pressing with a piece of metal until it was welded together. Once we had made the belt for our prototype, we tested it. The engagement to the rotation/position control wheel was slippery. After consultation with Dustin, he directed us to the use of timing belts. We moved back to the Mk1 and tested it on the new wheel that was accurate to the one on the actual field. We went to the drawing boards for the next idea to generate ideas on what would go on the robot. After some searching online, we found a combination of belts and sprockets that would work. The time spent waiting for the pars was used to CAD the model,
and used to help and teach wherever we could. Personally, I helped our other group members with their own ideas for the spinner design. When we got the parts for the spinner, we observed Muphy's Law taking full effect. We got the wrong belts, and even if we got the correct ones, the sprockets would have been useless anyway. The belts we got worked properly with the sprockets. It was a quick modification to the design and presto! We got the plates made on the CNC and got it assembled. The rest of the day was split among CAD, messing with the geometry of a wrist, woodworking, teaching chain, and finding a gear reduction that working with the NEO 550s.
We also worked closely with creating a wheel design along with a belt. We originally had the thought of having the wheel be directly plugged into the motor through a tube, but then realized it needs to be attached to the gearbox to reduce the speed of the fast motor. With some help, we created a design of an omni wheel on a gear box connected to a motor. Tim showed me that with the omni wheel, the robot can drive into the top of the spinner at any angle whereas a fully rubber wheel wouldn't allow for much margin of error. Additionally, a color sensor and camera aren't needed to rotate the wheel to the correct color. The encoder can be used to rotate the spinner based on inch conversion to number of ticks on the encoder, as well as use the arc of each color section on the wheel to get to the correct color through pressing a button. Finally, we then decided to go with a design that moved the spinner on the side with thicker wheels because it would allow for less mechanical movement on the robot - a piston could push it up and the thick wheels would account for the + or - 1/2 an inch of the height of the spinner. Our goals now are to continue thinking of new ways to integrate my mechanism to the rest of the robot.
Intake: We first looked at the variable that we would need to consider, and what we would need to be adjusted when prototyping (the speed, size and type of wheels/rollers, the range of the roller, compression of the ball and height off the ground). With these variables in mind, as well as designs from past years, we came up with three major ideas for the intake, over the bumper, through the bumper, and using an arcing backstop. We quickly decided that we could combine the arc with backstop and through the bumper groups because the essential mechanism was the same. After these two major designs were decided we went by making rigs. We split into two groups, each focusing on a different design. For the through the bumper we decided to use old side bumpers attached on top of the dolly to act as our “robot”. The rig would then be made of 80-20, to account for adjustability. For the over the bumper we decided to use the 2018 drive base that was not in use to attach the rig. This rig was also made with 80-20. The first major development was the decision to use plastic rollers instead of wheels on the over the bumper. This was because they were more consistent, robust, and would be easier to fix. For the through the bumper we looked at using small mecanum wheels to maneuver the ball to the center of the robot. We found that to active the wheels correctly they must be kept a maximum of an inch apart from each other as well as very close to the bumpers. Overall, we decided that the over the bumper mechanism would work better, because it focused on maneuvering the ball in the robot, rather than outside the robot where the ball could be lost. Now we are working with the hopper group and looking at a mechanism that can take the balls from the intake and move them with the balls jamming.
Shooter: This week, our group worked on the shooter for the robot. We tested multiple variables including launch velocity, launch angle, distance from the goal, wheel composition, compression, friction of the back plate, and the height of the shooter. Our group went through three prototypes this week. Our first one was just a roughly made flywheel with back plate using a shooter frame from a previous robot. Our second prototype was a more refined version with easier access to test wheel composition.
Our problem with our second prototype was that it was very difficult to accurately test compression without making some major changes. We then sat down with Zygmont and he took us through a new design for our third prototype, based off a shooter we saw on Westcoast Products. This prototype made it much easier to test compression because there are holes placed at various distances to move the back plate to yield the most desired compression. We are slowly making progress each day, and everyone is learning a lot about how to design shooters to test each different variable. We struggle figuring out the most logical way to put the shooter together, but we should improve with this the more we continue to build and learn. Zygmont has been very helpful with guiding us and helping us figure out what our next steps should be. Our goal for this next week is to test different gear ratios to test launch velocity and to hopefully find a way to make the angle adjustable. We are also going to build a base for the shooter because it is becoming too powerful to just hold still.
Hanger: We all got together and shared our opinions about what was feasible and what wasn’t worth it. We ruled out doing a buddy hang and spoke heavily about whether we wanted to do a hanger that could move laterally on the bar or not. We began by testing how the 2018 hanger hooks would work when holding onto the bar. When we found out that they would be almost perfect for the job we began by creating a mechanism that would get us to the bar. Our first iteration was to have one section of the arm that would fold out and the other part would telescope. Now we are planning on doing a double spring-loaded telescoping arm that is powered by a winch. Something else that we had prototyped was a laterally moving hook. We’ve tried different quantities of wheels and different spacing. The prototype we created allowed us to test different variables one at a time which was a focus of this year and we learned that adding two wheels to cradle the bar is better than one and we added delrin to the sides to reduce friction on the aluminum. We are currently working on upgrading our prototype to an L bar a person can sit on to allow us to get a more realistic representation of the robot hanging and the center of mass. When we tested it on an angle it had trouble and the wheels spun in place, but the hook was not vertical. It was angled to the side so most of the weight was not on the wheels. We are hoping that when the robot is on it would be less angled and most of the force of gravity would be on the wheels and not the delrin giving the wheels enough traction to drive. By making it a more realistic model we can accurately see if our idea is viable or whether if we need to do a redesign to distribute the weight differently so that the wheels have more traction. We spent a lot of this week researching extension arms and we have come up with an extremely simple idea that we believe will excel and we are going to spend the next week designing that and testing the prototype to see what we need to change next. Overall, we had a very successful week and have done very productive testing
Hopper: We tested ways to maneuver the power cell from point A to B. One of our first ideas were using belts to transfer the balls from the intake to the shooter in a path at least 35in in order to hold all five power cells. We thought of many ideas such as a horseshoe path or some with corners. With timing belts we found it effectively moves the ball through the path even around corners, but since these belts are sticky, we would have to index each ball coming it. Doing this would also stop the motion of all the balls in the robot. When we go to start it back up to shoot, it may take more time. Indexing the power cells would also be a lot more work for software as we would have to add sensors and different command groups. Because of this we wanted to build a hopper that is more continuous. Next, we decided to try a carrousel idea. Where the bottom spins from the inside, the outside works as a wall to keep the power cells in. However, we had the issue with getting balls in and out up to the shooter. Now we are trying a carousel that spins from the outside and the middle is a tube to bring the balls up to the shooter when wanted. We are currently CADing and building the prototype.
Drivetrain: Drivetrain was discussed in a team meeting, making a pro and cons list for both tank and swerve drive. Swerve drive would provide lateral movements in any directions on the field, but it is risky since the team hasn’t done it before, aside from experimentation with it during the offseason. Tank drive is reliable, and we know it works, but it doesn’t allow for movement in all directions like swerve can. We have not made a definitive yes or no answer, but it seems that they team is overall leaning towards a swerve drive robot.
Software: We continued working on getting our swerve code to work. After looking and deriving from some of 1323's swerve code, we got code to work. We continued to try and improve the code, implementing in a Pigeon Imu to work as a gyro. We are now trying to get more complex swerve code working. We are still working on this as of now.