-
-
Logo
Inspiration
Initially, we had planned on doing a completely different project related to sign language translation and ai recognition, but once we saw the air quality sensor, we had a different idea. Tens of thousands of people lose their lives to drunk driving, and every one of them matters. We want to be able to get that number to 0.
What it does
The ClearDrive sensor takes repeated measurements of the drivers breath using A Grove air quality sensor and gives back a value based on the concentration of foreign contaminants. Once that number reaches a certain limit, the car takes control. In our code, there is a time limit of when the car gives back control, which in real application will be much longer or infinite while the police are automatically called and help arrives.
How we built it
We built this using an Arduino UNO R4 Wifi, an Arduino shield to reduce soldering, a Grove 1.3 Air Quality Sensor, and a USB-C cable to transfer information. We used prompt engineering with OpenAI to develop the source code, and used the open source program CARLA to emulate driving in a simulation.
The Arduino reads the air quality through the sensor which is then sent to the computer. The computer reads that number in a python script, compares it to a hard coded value, and determines whether or not to enter self-driving mode. Once it enters self driving mode, the user is no longer given control and attempting to steer the car results in no change.
Challenges we ran into
We ran into MANY challenges along the way, most of all our inexperience in this area. Our main problem was hardware, as our computers can not handle the simulation effectively. If we were to do this again, it would be with a higher GPU PC to allow for more RAM usage. Also a struggle was learning the different functions that the Arduino and CARLA offer in python that can allow us to read data and use it effectively.
Another major issue is that the air sensor cannot differentiate between alcohol and poor quality air, rendering it useless in areas with very low air quality. If we were given proper funding, this issue could be solved.
Accomplishments that we're proud of
We are very proud to even have a running piece of code, but more than that an entire car simulation that responds to external stimuli. We learned a lot during this process, some about code but more about ourselves. We cannot wait to enter another hackathon in the future. Also, learning proper package installation was a good skillset to add, as that will allow us to become more independent coders in the future.
What we learned
We are happy to say we learned how to use the Arduino Shield during this process and found it fascinating the capabilities of the air quality sensor. We also learned a lot about prompt engineering, as we typically try to stay away from AI, but we found it very useful during this challenge.
What's next for ClearDrive
If we were to continue this project, we would find funding and try to sell this to companies like Tesla with self-driving cars. This could be a huge way to save lives now, and putting it off would only be a waste of human life.
Log in or sign up for Devpost to join the conversation.