Context-Aware Multiagent Broad Reinforcement Learning for Mixed Pedestrian-Vehicle Adaptive Traffic Light Control

0202 electrical engineering, electronic engineering, information engineering 02 engineering and technology
DOI: 10.1109/jiot.2022.3167029 Publication Date: 2022-04-13T19:36:53Z
ABSTRACT
Efficient traffic light control is a critical part of realizing smart transportation. In particular, deep reinforcement learning (DRL) algorithms that use neural networks (DNNs) have superior autonomous decision-making ability. Most existing work has applied DRL to lights intelligently. this article, we propose novel context-aware multiagent broad (CAMABRL) approach based on (BRL) for mixed pedestrian-vehicle adaptive (ATLC). CAMABRL exploits the system (BLS) established in flat network structure make decisions instead structure. Unlike previous works consider attributes vehicles, also takes states pedestrians waiting at intersection into consideration. Combining with mechanism utilizes adjacent agents and potential state information captured by long short-term memory (LSTM) network, can farsighted alleviate congestion. The experimental results show several state-of-the-art (MARL) methods.
SUPPLEMENTAL MATERIAL
Coming soon ....
REFERENCES (54)
CITATIONS (34)