deeptraffic and deep_traffic
The first is the official MIT competition platform while the second is a third-party submission that achieved top-ranking results within that competition, making them ecosystem siblings where one provides the framework and the other demonstrates a high-performing solution.
About deeptraffic
lexfridman/deeptraffic
DeepTraffic is a deep reinforcement learning competition, part of the MIT Deep Learning series.
DeepTraffic is a competition where you design and test a neural network to control autonomous vehicles on a simulated highway. You input your proposed network code, and the system simulates traffic flow, allowing you to visualize how your vehicle (or multiple vehicles) navigates dense traffic. This is for anyone interested in exploring how AI can optimize traffic flow and improve autonomous vehicle navigation.
About deep_traffic
gsurma/deep_traffic
MIT DeepTraffic top 2% solution (75.01 mph) 🚗.
This project offers a highly effective strategy for navigating simulated traffic using artificial intelligence. It takes in traffic conditions and vehicle behaviors within a simulated environment and produces optimal driving decisions to achieve high speeds without collisions. This would be used by researchers or students exploring advanced AI for autonomous vehicle control or traffic management.
Scores updated daily from GitHub, PyPI, and npm data. How scores work