Inspiration
In accordance with the competition theme of space, we quickly set our sights on a project that was in some way connected to remarkable systems we've seen in the lunar and mars rovers. Considering the time constraint and available resources, we soon narrowed down our interest to the various camera and remote sensing systems of these rovers, but particularly, the Perseverance. Ultimately, we based our final project on Perseverance's SuperCam system, which uses a combination of cameras, lasers, and other sensors to analyze rock and soil samples.
What it does
Our project utilizes an Arduino mega microcontroller, servo motor, and ultrasonic sensor to take distance readings across a 180 degree radius and examine patterns in the dataset to interpret the presence of objects and classify them by shape. More specifically, it sweeps the servo back and forth and notes the distance readings from the ultrasonic sensor. Then, it essentially analyzes changes in the distance readings to eventually recognize the difference between a flat surface, a rectangular prism, and sphere. In effect, this type of object recognition and classification is meant to represent a highly rudimentary and simplified version of the SuperCam system we were inspired by.
How we built it
Structurally, we designed and 3D printed all of the components, while still considering how the final design would be compatible with our hardware.
Challenges we ran into
In general, our group's programming experience was relatively limited, so we needed to do plenty of research and spend most of our time working through our program's logic. Even in its final stage, we're sure there is still plenty of room for improvement in this area.
Accomplishments that we're proud of
Overall, I think we are most proud of the fact that we managed to complete the project at all, but also that we were able to meet at least some of our initial goals. While our final product still has a somewhat long way to go if it is to reach the level of accuracy and polish we hoped for, we are certainly proud of the progress we managed to make over the quarter.
What we learned
Throughout the quarter, I'd say we made the biggest strides in terms of our familiarity and competence with programming in general. With the help of various mentors and through our own research, we were able to improve greatly at working with microcontrollers, prototyping circuits, and coding in Arduino.
What's next for TerraVision
There are actually plenty of interesting directions we can take to further advance our project. For example, we could implement a pan-tilt system to replace our more limited panning setup, we could design a more compact and comprehensive housing to completely enclose all of our hardware, we could implement buttons and LEDs to improve interaction and give clear indicators (i.e. object count, type), make our circuit more concise with soldering, refine our program, or experiment with more accurate sensors such as LIDAR. Additionally, we could also take the project a step further and use a more complex system based around for example an esp32cam and a machine learning algorithm, possibly through a platform like edge impulse. In fact, this was initially a loftier goal of ours, but we decided in the short term our more deterministic approach was more viable.
Log in or sign up for Devpost to join the conversation.