Open AccessReal‐time coordination of connected vehicles at intersections using graphical mixed integer optimization
Author(s) -
Ge Qiang,
Sun Qi,
Wang Zhen,
Li Shengbo Eben,
Gu Ziqing,
Zheng Sifa,
Liao Lyuchao
Publication year - 2021
Publication title -
iet intelligent transport systems
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.579
H-Index - 45
eISSN - 1751-9578
pISSN - 1751-956X
DOI - 10.1049/itr2.12061
Subject(s) - integer programming , mathematical optimization , robustness (evolution) , linear programming , benchmark (surveying) , computer science , computation , integer (computer science) , granularity , graph , algorithm , mathematics , theoretical computer science , programming language , biochemistry , chemistry , geodesy , gene , geography , operating system
Management of connected vehicles at unsignalised intersections is a large‐scale complex problem with safety constraints and time‐varying unsolved variables, which is crucial but hard to solve online. A faster coordination system, however, not only benefits from smaller time granularity to find optimum, but also has more robustness towards a scenario with fast‐moving vehicle nodes. This paper proposes a real‐time coordination scheme consisting of three stages. (a) Target velocity optimisation: collision‐free passage is formulated as a mixed integer linear programming problem, each approaching lane corresponding to an independent variable; (b) vehicle subgraph extraction: a directed graph is built and pruned based on the optimisation result, determining a subgraph wherein vehicle nodes pass without redundant time slot; (c) velocity profile synchronisation: velocity profile of the selected vehicles is planned synchronously, respecting inter‐subgraph constraints. The main contribution of this study is to propose a fast hierarchical optimization‐based coordination method, of which the complexity is invariant with the traffic density. Simulation has verified the effectiveness of the scheme from both microscopic behaviour and statistical characteristics, reducing single‐step computation time to 0.02 s, and saving average driving delay by 59.83% compared to the benchmark method.