Robo | Devpost

Inspiration

My team and I have been deeply involved with robotics throughout college and all of high school. We were introduced to these ideas through various organizations, and we believe that we can bring those skills and experiences to this datathon project.

What it does

In this project, we guided a circle equipped with multiple sensors through a line using sensor data. We adjusted the orientation based on this data, and from this, we implemented PID control, calculating both angular and linear acceleration.

How we built it

We implemented the PID Control System for the robot's movement adjustments. We also manipulated real-time inputs from the robot's sensors to adjust its movements. Integration was vital to ensure the robot stayed within its bounds; we didn't want the angular acceleration to exceed a specific angular velocity, which would make it spin out. It was also essential to initialize the robot by pointing at the line and start moving from there.

Challenges we ran into

We struggled with code related to rotations in circles, especially when the angular velocity got too high, which made it challenging to control the robot. Adjusting the PID values was a learning curve for our team. Furthermore, setting the robot's starting position was tricky as it wasn't always ideal.

Accomplishments that we're proud of

The code we developed enabled us to maneuver through the line and the graph seamlessly once we identified the starting point, showcasing the effectiveness of our algorithm.

What we learned

We delved deep into the workings of PID to calculate the angular acceleration and we also learned on how how to process and utilize data from robot sensor inputs. This experience also gave us insights into translating virtual robot adjustments to real-world applications. We also tried to use the buildup of erros to create limits and stop bindup of the robot so it doesn't get stuck in a loop

What's next for Robo

Next, we aim to enhance the robot's adaptability. We also want to refine the starting position of our robot. Currently, it occasionally starts off inaccurately, leading the robot to guess its orientation. We'd prefer to avoid such guesswork unless it's based on a trained model. Thus, improving the line-finding initiation process will be a priority.