Huan Vu'2,Samir Aknine'and Sarvapali RamchurnUniversite de Lyon.CNRS,Universite Lyon 1.LIRIS.UMR5205,Lyon 69622.FranceUniversity of Transport and Communications,Hanoi,Vietnam3 University of Southampton,Southampton,United KingdomAbstractfor human drivers,and givingtheoht-of.wconditions (about 60%reduced waiting time)anda market-based system where drivers have to purchase reser-1 Introductionmajor cities around the world suffer from high rates of trafficcongestion costs 6.9at major traAgainst this back ground,we propose a novel right-of-waybelieved that cars will be able to coordinate at intersections toto reachOptimiProblem (DCOP)which hasures of cars at all times.In a soped to optimise traftic flow at intersections.focusing on the[nstead,our approach focuses on discretisation of the inver--of-way527right-of-way allocation problem as follows.a set'=..where cach is the admission time■be the speed of the vehicle in cells per time stepat the[thiscle's priority).Thus our goal is tosearch for a minimisation:To ensure that the configuration satisties the mles de2 Problem Statement2(cf.Figure 1).This model is widely used in literature heretains the main proerties of s network while beingovertaking.We have:Thus,to minimise the travel time.we must minimise the waitt+,which is the carliest possible admission time of this+p.p is the time window for the waiting timeA small window may limit the search and03 DCOPs for Intersection Managementputational requirements for one agent.Morec=-mnlisation allows every agent to coordinate by exchanging3.2 Lanc-hased Approach[nstead of considering each vehicle as an agent.we can con-of thewhere variablethat has the knowledge of all vehicles in it.As a lane agent,DCOP depending on what agerts,variables and constraintsbased approach and a semi-deceutralised lane-based ap.thof ewhich03.1 Vchicle-based Approachhicles as agents.The number of agents is also the numberity of the solution to deal529