This summer I decided to create self-driving software during my holiday. The simplest way was to use some car simulator to get data and test my software. I searched for online car simulator games and used the first one that I'd found ( https://www.crazygames.com/game/russian-car-driver-zil-130 ). Then I recorded some training, validation, and test data taking screenshots and labeling them with pressed keys on the keyboard. The next step was the hardest, to design a neural network that would be able to predict how to behave in any situation on road. I had to take into account not only the quality of the predictions but also the simplicity of the network, so that computers could at once run game, and make predictions a few times a second. My first model was only able to return keys to press once per 3 seconds, which is definitely not enough. Finally, I got a model that was able to give quite satisfactory results in less than a second Then holidays ended and I had to focus on ot
read in google docs: https://docs.google.com/document/d/1gKWVnRc9ZYRUDh8GEUwlEUMuqMTSiP3DzFvezw82YHI/edit?usp=sharing Introduction to the problem Artificial intelligence and neural networks are really popular lately. I’m also interested in this topic and I work on many projects that use neural networks or machine learning. When I’ve first heard about CreateML I knew it will change things, and we will see more intelligent apps. It is so simple to have great results with this tool that I’m surprised that it’s not in more apps today. Then I thought about testing it. It’s great with simple tasks and it’s really easy to use, but the question is: Will it work with more complex tasks? To test it I need more complex tasks, so I need data that is harder to analyze. I’ve decided to work on photos of streets (I’m recording street while driving). Then there are steps that I will take: Train model to recognize maximum speed signs with CreateML Train model to recognize maximum speed
Komentarze
Prze艣lij komentarz