A game-theoretic framework for autonomous vehicles velocity control: Bridging microscopic differential games and macroscopic mean field games Article Swipe
YOU?
·
· 2020
· Open Access
·
· DOI: https://doi.org/10.3934/dcdsb.2020131
· OA: W2921029037
This paper proposes an efficient computational framework for longitudinal\nvelocity control of a large number of autonomous vehicles (AVs) and develops a\ntraffic flow theory for AVs. Instead of hypothesizing explicitly how AVs drive,\nour goal is to design future AVs as rational, utility-optimizing agents that\ncontinuously select optimal velocity over a period of planning horizon. With a\nlarge number of interacting AVs, this design problem can become computationally\nintractable. This paper aims to tackle such a challenge by employing mean field\napproximation and deriving a mean field game (MFG) as the limiting differential\ngame with an infinite number of agents. The proposed micro-macro model allows\none to define individuals on a microscopic level as utility-optimizing agents\nwhile translating rich microscopic behaviors to macroscopic models. Different\nfrom existing studies on the application of MFG to traffic flow models, the\npresent study offers a systematic framework to apply MFG to autonomous vehicle\nvelocity control. The MFG-based AV controller is shown to mitigate traffic jam\nfaster than the LWR-based controller. MFG also embodies classical traffic flow\nmodels with behavioral interpretation, thereby providing a new traffic flow\ntheory for AVs.\n
