BENTOgelion

Inspiration

Theme: Survival / D - Day The team wanted to build something that was at least somewhat realistic in a Doomsday setting: materials easily obtainable for the most part, and can help gather limited supplies.

What it does

BENTOgelion is a rover that scans a region for resources by travelling through the area as much as possible in an S-curve. BENTOgelion marks its location on the user interface when it detects objects such as food, water, or people. This allows the user to safely identify the location of resources without risking their own safety.

How we built it

Using the 300 credits that we were initially given, we quickly narrowed down the components of interest by considering our main objective of moving around the area to find resources. We thoroughly planned out how each component would interact with one another functionally and their logical locations on the chassis. Through work distribution, we divided the tasks into movement logic, computer vision, computer-aided design, and hardware integration. Through focusing on individual members' strengths and helping each other along the way, the design and build process was well time-managed, and the end product achieved satisfactory results.

Challenges we ran into

After initial attempts of putting components together to achieve the features we wanted, we encountered a few issues, such as component compatibility problems, modules simply not working, or modules performing under expectations. Through team discussion and decisions, we identified components to replace and ways of achieving a similar means to the function.
Due to Raspberry Pi 4 performance constraints, it was difficult to run a computer vision model locally. To address this issue, we decided to run Yolo on a local computer and stream it to the Pi using a local server.
Due to the weight placed on the motors by the portable charger, many times the hot glue connection that was keeping the motor on the chassis simply fell apart. In addition, since we were using the chassis plate provided by the Hardware Signout Spot instead of a custom one, we very often had to find creative uses of zip-ties, hotglue, and tape to secure components in place.

Accomplishments that we're proud of

Though using a computer vision model with the Raspberry Pi 4 was a painful setup process, we are extremely proud to have gotten it working with good efficiency, considering that using a Raspberry Pi is a brand-new experience to all of us.
Work was well-distributed and time was well-managed, so we have a working, presentable product hours before the submission deadline.

What we learned

Using "financial assets" wisely to create a low-cost yet effective product.
Working around hardware limitations such as underperforming modules and a lack of RAM on the Raspberry Pi 4.

What's next for BENTOgelion

According to Maslow's hierarchy of needs, love and belonging come after the basic survival needs that BENTOgelion aims to provide. For that reason, we hope to incorporate companionship into BENTOgelion as a potential next step by using large language models to interact with users vocally.