Swipe right in sequence while reading the caption.
Look at previous post before continuing.
Partner: @dan_ionov (made the machine learning library)
The simulation we have created mimics the evolutional theory that Charles Darwin had proposed. If you are curious about our algorithms or what is going on check out our github repository in my bio and read our report (6k+ words might be a very long read). ⠀⠀⠀⠀⠀⠀⠀⠀⠀
Slide 1: Neural network settings were found from weeks of testing the simulation.
Slide 2: In the beginning of the simulation, we force the car to learn how to drive straight and make a turn.
Slide 3: After tons of genetic mutations, the brain almost learns to make a turn.
Slide 4: The brain tries it’s hardest to beat the best fitness (fitness basically means score and is calculated by how fast the car reaches the max distance).
Slide 5: Learned how to do a full lap in 179 generations. The lowest Gen in my experience was 67.
Anyone is able to download this program in our github repository (called Builds.zip) and find the best car/AI settings themselves to get the least amount of generations. I do suggest reading our report first.
Music by: @l8nightprod
Next post will be in 1-2 weeks demonstrating the hardware and how our RC car is able to interpret the neural data to drive by itself.