- We began by attempting to compare the output of two different cognitive models- the microsoft cognitive API and our dataset source. We wrote code to do this, and hoped to find patterns, but eventually realized the impartiality of this project. The microsoft API allows only a very limited amount of free usage, and huge amounts of data would need to be transmitted over the internet- we were not confident about having the resources for this
- We therefore switched to a new line of thinking, and decided to address the following problem:
- Imagine you are a high school student walking through the halls trying to get to your locker. Suddenly, a crowd of people pass you from the opposite direction. Some avoid you, others collide into you, suddenly or suddenly swerve just before colliding. Imagine this now on a larger scale: THE ROAD Problem: How can we use the data and observations of an automated car to make roads safer by observing human behavior?
What it does
Use relative distance from the camera of an artificially intelligent car to objects that come into interaction with it in order to observe driver and pedestrian habits.
Challenges we ran into
Lack of time. Lack of sleep. And having a limited knowledge of DBMS/ file IO.
Accomplishments that we're proud of
- Defining a precise problem statement
- Establishing, clearly, what is required to solve it
- Writing part of the necessary code
- Performing basic mathematical operations to conclude about things like shutter speed to demonstrate that drawing conclusions of the nature we intend to is feasible.
What we learned
Handling datasets and how to draw useful conclusions from numbers
What's next for Automated cars
Completing the code, and refining to reach other useful conclusions