: The authors introduce a decentralized training method with centralized execution that handles the large, dynamic scale of urban transport networks.
The filename is the identifier for the supplementary code and data associated with the research paper "Learning to Control Autonomous Fleets via Sample-Efficient Deep Reinforcement Learning" . Paper Overview
: Learning to Control Autonomous Fleets via Sample-Efficient Deep Reinforcement Learning M_S_2o_6_k3gn.zip
The .zip file contains the of the algorithms discussed in the paper. The research focuses on:
: A novel Deep Reinforcement Learning (DRL) approach that uses a hierarchical structure to improve "sample efficiency," meaning the system learns effective strategies using significantly less data than traditional methods. : The authors introduce a decentralized training method
: Originally published in Proceedings of the 20th International Conference on Autonomous Agents and Multiagent Systems (AAMAS 2021) . Context of the File
: Filippos Christianos, Georgios Papoudakis, Aris Filos, and Stefano V. Albrecht. The research focuses on: : A novel Deep
: Optimizing the dispatching and rebalancing of autonomous vehicle fleets (e.g., ride-sharing services) to minimize wait times and maximize efficiency.