Challenges
Bearings of Catapult
One of the main challenges with the catapult design was finding ways to minimize resistance during the catapult launching phase. This included finding a way to disconnect the motor that is responsible for driving the catapult into its launch angle. Our initial designs utilized clutch bearings to allow torque transmission in one direction while allowing free rotation in the opposite direction. Upon acquisition of the clutch bearings though, we discovered that implementing this design would be more difficult than anticipated due to the fact that the force between the motor transmission and the axle never actually changes direction throughout the entire loading and launching cycle.
To solve this problem, we decided on a very simple solution that takes advantage of the fact that the catapult axle never rotates more than 120. The larger sprocket from the motor transmission system was left loose on the axle so that it could spin freely without also rotating the axle. A stainless steel pin was press-fitted into the face of the sprocket and, similarly, another pin was fixed on the end of the axle so that it extended radially from the axle's center of rotation. With this system, the motor could drive the catapult downward through the interaction of these two pins. When the catapult reached the desired launch angle, the brakes system could lock the axle in place, allowing the motor to spin the sprocket in reverse without the axle rotating. Since the axle never rotates more than 120, rotating the sprocket half a revolution eliminated any interaction between the motor and catapult axle during the launch phase.
Calibration
We found that the potentiometers had the tendency to drift and float between trials. To reduce the potential for compounding errors, we added a calibration protocol that runs at every startup to set the bounds of both the catapult arm and base swivel. With the bounds selected, we interpolated between those values to determine the intermediate pull-back angles and launch angles.
A variety of sensors and limiter microswitches were used in determining the position of the various BroBot subsystems. Two spring loaded microswitches are located at the top and bottom of the retrieval system, to mark the limits of the carrier elevator. Another microswitch is located below the brake lever arm to determine its state – whether the hydraulic brake is engaging the brake rotor or the catapult arm is free to swing. Two (one single turn and one multiturn) potentiometers were used as low-level encoders to determine the angular placement of the base and catapult arm.
Retrieval Springs
The springs proved to be much more complex to work with than we had anticipated. First of all, it was difficult to acquire the springs that we needed because we needed springs that could compress 2.75” so that the can could roll out of the cavity in the side of the fridge, but we also needed springs that would compress fully under a small load. Most off the shelf springs that could compress this much required very larges compressive forces. Therefore, after a few calculations, we decided to use two springs in series and in parallel that could then better support the weight of the fridge cover and still be able to compress a full 2.75” with smaller forces.
The original design for the retrieval springs and fridge cover was to mount the springs from a plate mounted on the bottom of the fridge. The other side of the springs would then be connected to the fridge cover. This however, proved to be a problem because the springs started to bend rather than compress under compressive forces. We tried adding metal spacers in the diameter of the spring and holes through the fridge cover so that the metal spacers served as guides for the fridge carries as the springs below it compressed. Although aspects of this design worked well, we realized that the fridge cover itself was interfering with the drive shaft once compressed halfway, and the springs began to bind due to the friction and resistance of the metal spacers.
To overcome these issues, we redesigned the fridge cover and decreased the length of the tabs that would support the can carrier, ensuring it would clear the drive shaft. We also realized that the springs fit perfectly inside the groove of the 8020 which could serve as a natural guide for the springs but only had a few points of contact. This reduced the friction allowing the springs to compress more easily while preventing them from bending or moving in any unwanted directions. This new design uses less material, is a more reliable and robust design and is even more aesthetically pleasing!
Lazy Susans
For the rotating base we decided to go with a lazy susan, a rotating ball bearing tray. We chose this because it was an off the shelf part that satisfied our needs for the base. The lazy susan we purchased was rated to a load of 1000 pounds. It performed well until we started test firing the catapult. When the catapult arm would hit the hard stop after release, the momentum of the catapult pried apart the two connecting plates. The material yielded on one side of the lazy susan and the two rotating plates jammed on each other. We ended up having to replace the lazy susan. We realized that the lazy susan wasn't designed to take forces pulling the two plates apart in tension; the load rating was for compression. In order to resolve this issue without a major redesign of the catapult we decided to preload the lazy susan using a bolt. The bolt was placed through the center of the lazy susan and the top of the refrigerator. The head of the bolt rested in the inner race of the bearing an the outer race rested on the top of the lazy susan. This allowed the lazy susan to rotate freely while applying a compressive load. The other end of the bolt rested on the inside of the fridge. With this new setup we did not encounter any further problems with the lazy susan.
Wiimote and Arduino Interface
Throughout the course of the project, there were a few challenges electrically. The Wiimote to Arduino interface was very difficult because there was not much support for it online due to its complexity. In addition, the code used for the USB host shield only supports one special Bluetooth dongle so there was a lot of trial and error in selecting the right dongle.
Fuses
We also had a few electrical difficulties that arose while integrating the electronics with the mechanical system. Power and ground from the power supply was accidentally reversed and destroyed many of our electronic components. After rebuilding and being extremely careful, we accidentally blew a fuse on our portable power supply. This issue was easily fixed by replacing the fuse.
Old Retrieval Design
New Retrieval Design