FISHBOT FRENZY

Ramping Things Up

This phase was focused on enhancing functionality and aesthetics of different parts of our project, along with limited amounts of integration.

Mechanical

The mechanical subteam's goal was to increase the complexity of the tank and score box, and to further iterate upon the fishbot shells.

This iteration of the tank was designed to be multilevel, with a DC motor rotating a central platform. Due to a lack of testing, the ramp was too steep and the upper level was too narrow for the fishbots to navigate; however, as a proof of concept of complexity and potential for a tiered tank design, it was successful!

The scorebox design was changed to mimic a large fish consuming the fishbot. While the design did not include the tunnel that would lead the fish back to the main tank nor any object detection, it was a successful proof of concept for a geared chomping mechanism. An SG90 Micro Servo Motor drove the spur gears that moved the jaws.

Finally, for the fishbots, we were able to create three different fishbots! We tested out different body shapes and different materials, including PLA and TPU. A eucaboard topper for aesthetics was considered but decided against due to the increased weight.

Electrical/Software

The main goal for electrical and software during this phase was to test RFID sensing. Unlike a simple distance sensor, RFID would allow us to assign different points to unique fish bots.

The software successfully integrated with the scorebox; the opening and closing jaws were limited in software rather than in hardware. Electrical also began working directly with mechanical to improve fishbot wiring.

Next Steps

Many challenges presented themselves during this phase. The tank needed to be rethought; the ramp didn't work, the paths were too narrow, and the square shape didn't work particularly well. The scorebox didn't fit any fish inside, had no tank connection (tunnel) designed, and had no plan for integrating with RFID. Our next steps included:

1. Making a tunnel to connect the score box and tank
2. Planning the RFID sensor mount
3. Integrating a score display outside of the Arduino IDE Serial Monitor Output
4. Determine how the score box would open and close

Phase 3: It's Beginning to Look a Lot like Fab Time

For this phase, we needed to finalize designs for a clear path to fabrication and integration.

Mechanical

The final iteration of the fish tank began to take shape in CAD. The size of the tank increased substantially from the previous phase and while the floor-moving mechanism was removed, three DC motors powering different obstacle mechanisms (scotch yoke and two spinners) were placed instead.

In order to allow access to the bottom layer of electronics, the structure was planned out such that the main platform could be removed along with two structs and half of the railings. Using six struts not only meant that four struts could be permanently in place, but also that the scoreboxes could be permanent attachments directly across from each other. The dome shape was partially for aesthetics, and partially to improve game balance; the added difficulty of fishing through a roof was expected to alleviate the fact that the fish were highly magnetic.

Planning for the score display started during this phase. The initial design located the score display on the back of the scoring boxes; however, this would prevent the opponent from seeing the scores. We chose to move the score display away from the score box and planned to mount them to the tank struts instead.

The score box was also initially intended to close in response to the placement of a fish using a distance sensor. However, implementing the distance sensor in such a cramped space with moving parts in a manner that wouldn't accidentally trigger in response to the box but would still detect fish dropped in. Ultimately, we chose to instead randomize the opening and closing mechanism of the scorebox, much like other popular magnetized tabletop fishing games, which would simplify the sensor layout.

The fishbot iteration hadn't ended yet, either! At this point, the team determined that the best course of action would be to design an internal PLA shell with an external TPU casing, rather than continue forward with the full TPU shell. A shift from 3V coin cell batteries to smaller 1.5V button cell batteries to save space and weight was made after determining that there was no noticeable speed shift, but this meant there needed to be a full overhaul of the CAD. For this sprint, no new shells were created, but rigorous testing was done to determine the next steps.

Electrical and Software

The focus for electrical and software was on continuing to work with RFID functionality, now integrating it with a seven-segment LED display to show points. Each time an RFID token passed in front of the sensor, the LED display would change its display to increment a number.

Final Steps

We now have the overall structure for how the game is going to be integrated between subteams. From here, our main focus is shifting to fabrication and iterative development to build, test, and refine each of our components before the final demonstration day!

Final Mile

Like many engineering projects with a deadline, we got the most work done during the final three weeks of our project! Here's the breakdown on how we flat-out sprinted through the rest of this game.

Mechanical

The tank was mostly done at this point; the scotch yoke, base, motors, and associated supports were all placed and dimensioned in CAD. The main components added during the final segment were the dome cap, a third layer of railings, and the 3D-printed obstacle mounts for the motors and scotch yoke.

However, it still wasn't finished enough to be cut and assembled immediately. The bottom layer of the railing design changed after testing determined that the fishbots would get stuck on sharp corners; a third layer was added to increase support and adhesive contact to the struts, and then altered along with the second layer in response to the scorebox location and design.

The 3D prints required some level of iteration as well. They were designed with pressfitting in mind, as they would need to be removed to slide the main platform off, and as such they required testing until the tolerances were acceptable. Likewise, the struts were originally meant to be made of clear acrylic for aesthetic purposes, and so was made to be pressfit. Difficulties with laser printing, tight tolerancing, and lack of stock meant that we were forced to use plywood as a replacement in the end.

For the scorebox, a large concern was our inability to control the orientation that the fishbots fell. An initial concept involved preventing them from flipping by designing the platform in a specific manner, such as through a ramp or a vertical tunnel.

Unfortunately, this method did not succeed, and we attempted to instead design the fishbots to consistently land upright. With this in mind, the platform simplified drastically and became much more focused on funneling the fishbots out into the tunnel.

This design, however, was also problematic; the assembly was never completed in favor of being able to test the railing and platform faster, and it was unexpected when the fishbots were caught on the bottom edges of spur gears and the fish jaws.

Finally, due to time constraints, we resorted to removing the "eating" mechanism.

Throughout all of this design and redesign, we were also fabricating parts. The main platform was cut with the Shopbot CNC Router with little issue. We had significant issues with cutting with the laser cutters, where there were errors with speed levels and poor tolerancing due to settings and overuse of the shop spaces. Pieces that didn't cut all the way through were difficult to retrieve and happened often enough for it to be a time issue.

This last phase included two fishbot shell versions, each of which had multiple subiterations. Shell 7 was a PLA-core bot with TPU legs designed to hold two 1.5V button cell batteries.

However, this shell resulted in a fishbot that was too tall. While it was high enough to be read by the RFID sensor, it would get stuck on the ¼" plywood railings. The next iteration involved lowering the face of the fishbot and adding a pointed tip to turn it away from walls instead of running directly into them.

The other benefit of shell #7b is that it has a lower center of mass. We had hoped this would encourage it to land feet-first more often. It did not. In order to solve the problem of orientation when landing, a different idea was devised: making the fishbots double-sided.

The TPU legs of Fishbot 8 clip on to the center and have legs on both sides, allowing the fish to move regardless of which side it lands on.

Electrical and Software

• circuitry finalization
  • all prototype circuits were converted into soldered perfboard ones
• battery power
  • each arduino was tested to ensure it was able to be powered from a 9V battery
• 7 segment displays
  • a second digit was added to each scorebox for a total of two digits per scorebox
• RFID testing
  • final tests were performed to double check the range of the RFID sensor - we found it could reliably detect chips from 2-3cm away