Fuzz-Bee (2018)
Fuzz-Bee is an AI race car application developed to test the efficiency of a fuzzy logic AI racer (violet rectangle) with a finite state machine AI racer (red rectangle), in following the race track line. (Assumption: Linear velocity is constant).
Goal of the project
The goal of Fuzz-Bee was to implement a racing bot that would track the race line. To keep the project scope simple, it was assumed that the linear velocity of the bot's race car would remain constant. This project provided valuable insights into the varying reactions of different AI models in a given situation.
Why this project?
I opted for Java and the Processing engine to display a basic line and two rectangles symbolizing AI models: Finite State Machine and Fuzzy Logic. The project aimed to enable the user to shift the racing line using the left/right keys and observe the AI models in action, followed by a detailed analysis of their performance. The main goal was to gain a comprehensive insight into the behavior of AI in specific scenarios.
What were the problems faced during the development of this project?
Utilizing MATLAB for generating the fuzzy logic values proved to be challenging due to the learning curve associated with navigating the software.
How was the problem resolved?
To keep things simple, I had a simple right angled triangle approach to the calculation of the steer. The calculation happened every frame with the AI deciding to steer towards to away from the race line based on how far it is from it
