Accepted author manuscript, 2.02 MB, PDF document
Available under license: CC BY: Creative Commons Attribution 4.0 International License
Final published version
Research output: Contribution to Journal/Magazine › Journal article › peer-review
Research output: Contribution to Journal/Magazine › Journal article › peer-review
}
TY - JOUR
T1 - Congestion-Aware Path Planning With Vehicle-Road Cooperation in AIoV
AU - Liu, Guoqiang
AU - Bilal, Muhammad
AU - Wang, Qingyang
AU - Xu, Xiaolong
PY - 2024/12/15
Y1 - 2024/12/15
N2 - With the gradually integration of Internet of Vehicles (IoV) and Artificial Intelligence (AI), Artificial Intelligence of Vehicles (AIoV) is emerging as a novel paradigm with advanced capability for information gathering and decision-making. Leveraging massive traffic information facilitated by vehicle-road coordination in AIoV, path planning has the potential to effectively mitigate existing traffic problems, such as road congestion, improving traffic performance. However, the dynamic nature of traffic flow and the complexity of road networks increase the difficulty of path planning, posing a serious threat to road safety. In response to this challenge, a reinforcement learning based path planning scheme with traffic flow prediction, named RPFP, is proposed. RPFP consists of two fundamental components: precise traffic flow prediction and intelligent path planning. Specifically, the temporal convolutional network (TCN) is innovatively integrated into the spatiotemporal graph neural network (STGNN), providing accurate traffic flow prediction by comprehensively capturing spatial and temporal patterns. Informed by predicted traffic congestion, a path planning method utilizing dueling double deep q-network (D3QN) algorithm is employed to navigate within complex road networks. Eventually, RPFP was evaluated for its effectiveness through comprehensive experiments conducted on real traffic datasets. The superiority of RPFP was further substantiated via comparisons with multiple baseline schemes.
AB - With the gradually integration of Internet of Vehicles (IoV) and Artificial Intelligence (AI), Artificial Intelligence of Vehicles (AIoV) is emerging as a novel paradigm with advanced capability for information gathering and decision-making. Leveraging massive traffic information facilitated by vehicle-road coordination in AIoV, path planning has the potential to effectively mitigate existing traffic problems, such as road congestion, improving traffic performance. However, the dynamic nature of traffic flow and the complexity of road networks increase the difficulty of path planning, posing a serious threat to road safety. In response to this challenge, a reinforcement learning based path planning scheme with traffic flow prediction, named RPFP, is proposed. RPFP consists of two fundamental components: precise traffic flow prediction and intelligent path planning. Specifically, the temporal convolutional network (TCN) is innovatively integrated into the spatiotemporal graph neural network (STGNN), providing accurate traffic flow prediction by comprehensively capturing spatial and temporal patterns. Informed by predicted traffic congestion, a path planning method utilizing dueling double deep q-network (D3QN) algorithm is employed to navigate within complex road networks. Eventually, RPFP was evaluated for its effectiveness through comprehensive experiments conducted on real traffic datasets. The superiority of RPFP was further substantiated via comparisons with multiple baseline schemes.
U2 - 10.1109/jiot.2024.3446699
DO - 10.1109/jiot.2024.3446699
M3 - Journal article
VL - 11
SP - 39629
EP - 39636
JO - IEEE Internet of Things Journal
JF - IEEE Internet of Things Journal
SN - 2327-4662
IS - 24
ER -