video-game-models

CARLA: An Open Urban Driving Simulator

CARLA is an open-source simulator designed for the development, training, and validation of autonomous urban driving systems. It is an excellent tool for researchers and developers to test their ideas regarding self-driving cars, with the goal of improving the safety and functionality of autonomous vehicles. The Development of CARLA CARLA was developed from the ground up to support the needs of researchers and developers working on autonomous driving systems. The simulator includes open-sourc

DeepMind AlphaStar

AlphaStar is an advanced reinforcement learning agent designed to tackle the challenging game of Starcraft II. It uses a policy that is learned through a neural network with various types of architecture, including a Transformer for processing observations of player and enemy units, a core LSTM for handling temporal sequences, and a Residual Network for extracting minimap features. To manage the combinatorial action space, AlphaStar uses an autoregressive policy and a recurrent pointer network.

1 / 1