This project uses an algorithm I came up with. I based it off of a algorithm called a Sigmoid Neural Network, but this is not a Sigmoid Neural Network. I have set up 4 variables. Hesi, (hesitation), time, best hesi, and best time. If you choose to train a new model, it sets hesi to a random integer between 0.1 and 2. It waits (hesi) seconds and then jumps over the laser. The cloud system detects what the highest time is and the hesi value which resulted in that time. The best hesi value is used on the 'trained model' , which is activated when you press two. That is why a trained model will last longer than the raw, untrained model. I would recommend trying the untrained model a few times, because there is a chance that you could get a lucky hesi, (similar to a weight in a Sigmoid Neural Network), and your hesi will become the new highest hesi. Most of the time, it will end quickly because the hesi scores are very similar. I call it a refining algorithm because It refines itself based on the highest score. The goal of the bot is to dodge the laser.
