Loading [MathJax]/jax/output/SVG/jax.js
Research article Special Issues

Mean-field limit of a hybrid system for multi-lane car-truck traffic

  • In the present work we model multi-lane traffic flow in presence of two population of vehicles: cars and trucks. We first develop a finite-dimensional hybrid system which rely on continuous Bando-Follow-the-Leader dynamics coupled with discrete events motivated by the lane-change maneuvers. Then we rigorously prove that the mean-field limit is given by a system of Vlasov-type PDE with source terms generated by the lane-change maneuvers of the human-driven vehicles.

    Citation: Maria Teresa Chiri, Xiaoqian Gong, Benedetto Piccoli. Mean-field limit of a hybrid system for multi-lane car-truck traffic[J]. Networks and Heterogeneous Media, 2023, 18(2): 723-752. doi: 10.3934/nhm.2023031

    Related Papers:

    [1] Cécile Appert-Rolland, Pierre Degond, Sébastien Motsch . Two-way multi-lane traffic model for pedestrians in corridors. Networks and Heterogeneous Media, 2011, 6(3): 351-381. doi: 10.3934/nhm.2011.6.351
    [2] Matteo Piu, Gabriella Puppo . Stability analysis of microscopic models for traffic flow with lane changing. Networks and Heterogeneous Media, 2022, 17(4): 495-518. doi: 10.3934/nhm.2022006
    [3] Amaury Hayat, Benedetto Piccoli, Shengquan Xiang . Stability of multi-population traffic flows. Networks and Heterogeneous Media, 2023, 18(2): 877-905. doi: 10.3934/nhm.2023038
    [4] Carlos F. Daganzo . On the variational theory of traffic flow: well-posedness, duality and applications. Networks and Heterogeneous Media, 2006, 1(4): 601-619. doi: 10.3934/nhm.2006.1.601
    [5] Michael Herty, Adrian Fazekas, Giuseppe Visconti . A two-dimensional data-driven model for traffic flow on highways. Networks and Heterogeneous Media, 2018, 13(2): 217-240. doi: 10.3934/nhm.2018010
    [6] Dirk Helbing, Jan Siegmeier, Stefan Lämmer . Self-organized network flows. Networks and Heterogeneous Media, 2007, 2(2): 193-210. doi: 10.3934/nhm.2007.2.193
    [7] Paola Goatin, Sheila Scialanga . Well-posedness and finite volume approximations of the LWR traffic flow model with non-local velocity. Networks and Heterogeneous Media, 2016, 11(1): 107-121. doi: 10.3934/nhm.2016.11.107
    [8] Emiliano Cristiani, Smita Sahu . On the micro-to-macro limit for first-order traffic flow models on networks. Networks and Heterogeneous Media, 2016, 11(3): 395-413. doi: 10.3934/nhm.2016002
    [9] Mauro Garavello . A review of conservation laws on networks. Networks and Heterogeneous Media, 2010, 5(3): 565-581. doi: 10.3934/nhm.2010.5.565
    [10] Benjamin Seibold, Morris R. Flynn, Aslan R. Kasimov, Rodolfo R. Rosales . Constructing set-valued fundamental diagrams from Jamiton solutions in second order traffic models. Networks and Heterogeneous Media, 2013, 8(3): 745-772. doi: 10.3934/nhm.2013.8.745
  • In the present work we model multi-lane traffic flow in presence of two population of vehicles: cars and trucks. We first develop a finite-dimensional hybrid system which rely on continuous Bando-Follow-the-Leader dynamics coupled with discrete events motivated by the lane-change maneuvers. Then we rigorously prove that the mean-field limit is given by a system of Vlasov-type PDE with source terms generated by the lane-change maneuvers of the human-driven vehicles.



    Mathematical models for traffic flow are mainly classified into microscopic, mesoscopic, macroscopic, and cellular, depending on the scale at which they represent vehicular traffic [1,4,39]. Generally, the scale is chosen according to the type of traffic characteristics to be captured. In this paper, we are interested in microscopic models and mesoscopic descriptions.

    Microscopic models are developed with the idea of explicitly reproducing the individual behaviors of drivers, such as reactions to traffic changes and interactions with other vehicles, therefore the dynamic is expressed in terms of trajectories of the single vehicles, by means of ODEs. Two of the most successful microscopic models are the Optimal Velocity model, also known as the Bando model [3], and the Follow-the-Leader model [20,27,43], in which the acceleration of the single vehicle is controlled according to the velocity of the leading vehicle.

    Mesoscopic traffic flow models were derived as bridge between the family of microscopic models and the family of macroscopic models which interpret traffic as a continuum flow. Usually mesoscopic models describe vehicle flow in aggregate terms such as in probability distributions. Mean-field equations fall into this category and aim to provide an aggregate and statistical view of traffic by capturing and predicting the main phenomenology of microscopic dynamics. Among the relevant literature, in this context, we can mention classic works like [35,40,41] and more recent results, e.g., [11,15,27,28,34,36]. The passage from microscopic to mesoscopic description can be also rigorously performed by using the generalized version of the classical Wasserstain distance [2]. The analysis, through the progressive change of scale, is not only a peculiarity of traffic flow models, but extends to other research areas such as biology [12,14], economics [48] and social sciences [13].

    Heterogeneous and multi-lane traffic flow modeling is fundamental to understand the dynamics and control of complex traffic systems. Specifically in this work we consider two populations of vehicles: cars and trucks. For other relevant contributions in multipopulation traffic models, see [5,9,32,42,49]. We model the multi-lane traffic by hybrid systems because of its hybrid nature: the continuous dynamics on each lane and the discrete events due to lane-changing maneuvers. The lane-change is one of the most common maneuvers, which generates interaction and risk [29] among vehicles on motorways. Current models for multi-lane traffic include two-dimensional models [26,46], in which lane changing rules are not explicitly prescribed and models treating lanes as discrete entities [7,30,44].

    Our main contribution regards the formalization of the passage from microscopic dynamics to mesoscopic in the case of the two before mentioned populations of vehicles (cars and trucks). The model used is the combined Bando-Follow-the-Leader one for both the populations. In particular, we reformulate it by replacing the interaction with the closest vehicle ahead with a short-range interaction kernel which allows to write the system of ODEs in a convolution framework. Continuous dynamics are combined with discrete dynamics generated by the lane changing rules, which are designed following [33]. This leads to an hybrid system (see [8,19,21,37,47]) whose mean field limit is given by a system of two Vlasov-type PDEs with source terms [17,25,31]. These source terms are generated by the discrete lane changing rules and induce the measure solutions to change mass in time, thus the limit is obtained using the generalized Wasserstein distance [38].

    This complete representation of multi-lane heterogenous traffic at microscopic and mesoscopic scales together connected by a rigorous limiting procedure has also been extendended by the same authors to the case of two populations of human-driven vehicles and autonomous vehicles [10]. The main difference is that a control is introduced in the acceleration of autonomous vehicles with the idea that they can influence the general dynamics of the other two populations. Moreover the number of autonomous vehicles remains finite in the limiting procedure.

    The paper is organized as follows. In Section 2 we introduce the notation used to capture the heterogeneous traffic and the main notions necessary for what we are going to prove. We describe in detail the combined Bando-Follow-the-leader model and how to reformulate it in convolution form. This is propaedeutic to the derivation of the mean-field limit. The lane change rules, together with the key ideas behind, are explored in the Subsection 2.2. Soon after we present an overview on the generalized Wasserstein distance and a revised version of Ascoli-Arzelá theorem which play an essential role in the process of derivation of the Vlasov-Poisson type equations with source term. In Section 3 we define the "cool-down" time model assumption, which is critical to describe the frequencies of the vehicles' lane-changing behavior and to prove the well-posedness of our heterogeneus multi-lane traffic model. Then we introduce the finite dimensional hybrid system which captures the continuous dynamics on each lane and discrete events for lane-changes. Section 4 is finally devoted to the rigorous derivation of the mean-field limit for the hybrid system which leads to two coupled Vlasov Poisson type equation with source term. In Section 5 we discuss future research directions opened by this work.

    In this section, we recall the original and convolutional form of a car-following model, Bando-Follow-the-leader model, and design lane-changing rules based on distance headway and acceleration for multi-lane traffic in both homogeneous and heterogeneous traffic conditions. We also give an overview on the generalized Wasserstein distance [38] and a revised version of Ascoli-Arzela theorem introduced in [23]. In the end, we formally derive the mean-field limit of a finite dimensional system and listed the results on partial differential equations of Vlasov-type with and without source terms.

    The Bando-Follow-the-Leader (Bando-FtL) model is a first order car-following model introduced in [16]. The main idea of the Bando-FtL model is that the ego vehicle adjusts its own acceleration based on its space headway, optimal velocity (determined by its space headway) and its leader's velocity. We refer the readers to [22] for the well-posedness of the Bando-FtL model.

    We assume that the vehicles drive in the same direction and vehicle n+1N+ is the leader of vehicle nN+. Let (xn,vn):[0,T]R×R0 be the position-velocity vector of vehicle n, where T>0 is fixed, lnR>0 be the length of vehicle n, hn=xn+1xnln be the space headway of vehicle n, and V:(0,+)[0,+);hV(h) be the optimal velocity function which describes the desired velocity corresponding to space headway. Usually, the optimal velocity function V is increasing with respect to the headway. For example, one may choose the following optimal velocity function as in [45], for any h(0,+),

    V(h)=vmaxtanh(hds)+tanh(lv+ds)1+tanh(lv+ds), (2.1)

    where lv is the vehicle length and ds is the minimum safety distance required in the model. We recall the Bando-FtL model in two traffic conditions: homogeneous and heterogeneous. In the case of homogeneous traffic where the vehicles' physical dimensions do not vary much, the governing equation of the Bando-FtL model is as follows: for vehicle nN+

    {˙xn=vn,˙vn=α(V(hn)vn)+βvn+1vn(hn)2, (2.2)

    where α,β are positive with proper dimensions. We develop the homogeneous Bando-FtL model to its heterogeneous form by giving subscripts to the model parameters as follows: for vehicle nN+

    {˙xn=vn,˙vn=αn(Vn(hn)vn)+βnvn+1vn(hn)2, (2.3)

    where αn,βn are again positive parameters with proper dimensions, Vn is an optimal velocity function depending on the headway hn=xn+1xnln. In the case of heterogeneous traffic with cars and trucks, due to the four different car-truck car-following combinations (C-C, C-T, T-C, T-T), all parameters αn, βn and the optimal velocity function Vn have four different alternatives. For example, αn can be αcc, αct, αtc, or αtt. Here αct represents the weight of the Bando term in the Bando-FtL model in the case of car-following-truck. The vehicle length ln has two alternatives the car length lc>0 and the truck length lt>0.

    Now we want to rewrite the heterogeneous Bando-FtL model (2.3) into its convolutional form. Instead of only considering one leading vehicle ahead, the drivers adjust their accelerations and decelerations according to the types and velocities of vehicles in nearby front, their own velocities and the optimal velocities depending on their space headways. Of course, we cannot expect that the strength of the interaction in C-C case is the same as in T-T case, and also the order of two different type of vehicles plays an important role. Hence the strength of the interaction in the configuration car-truck must be different based on the car-truck car-following combinations. Therefore, we assume that the ego vehicle only interact with other front nearby vehicles that is at most εn>0, n{cc,tc,ct,tt}, away. We call εn the strength of interaction.

    For convenience of notation, for the heterogeneous traffic containing cars and trucks, we introduce I={1,,P+S} be the set of index for all the vehicles, IP={1,,P} the set of index for cars and IS={P+1,,P+S} for trucks. Note that I=IPIS. We define the following two time dependent atomic probability measures

    μP(t)=1PiIPδ(xi(t),vi(t)),μS(t)=1SiISδ(xi(t),vi(t)) (2.4)

    tracking the position-velocity of cars and trucks at time t[0,T].

    Consider convolution kernels of the form

    Hn1:R×R0R with n{cc,tc,ct,tt}(x,v)αnhn(x)(Vn(x)v)

    where hn:RR0 is a smooth function supported compactly on [εn,0]. Here hn measures the interaction of two vehicles depending on their distance and types. The Bando-Term in Eq (2.3) can be rewritten as

    (Hcc11μP+Htc11μS)(xi,vi)=αccPkIPhcc(xixk)(Vcc(xkxi)vi)+αtcSkIShtc(xixk)(Vtc(xkxi)vi) (2.5)

    for cars (iIP) and

    (Hct11μP+Htt11μS)(xi,vi)=αctPkIPhct(xixk)(Vct(xkxi)vi)+αttSkIShtt(xixk)(Vtt(xkxi)vi) (2.6)

    for trucks (iIS). Here 1 is the convolution with respect to the first variable.

    In analogous way we introduce four kernels

    Hn2:R×R0R with n{cc,tc,ct,tt}(x,v)βnhn(x)vx2

    and rewrite the FtL-term in Eq (2.3) as

    (Hcc2μP+Htc2μS)(xi,vi)=βccPkIPhcc(xixk)vkvi(xixk)2+βtcSkIShtc(xixk)vkvi(xixk)2, (2.7)

    for iIP and

    (Hct2μP+Htt2μS)(xi,vi)=βctPkIPhct(xixk)vkvi(xixk)2+βttSkIShtt(xixk)vkvi(xixk)2, (2.8)

    for iIS. In this case is the convolution with respect to both space and speed.

    This leads to the following convolutional formulation of the Bando-FtL model with two distinct dynamics for cars and trucks:

    ˙xi=viiI˙vi={(Hcc11μP+Htc11μS)(xi,vi)+(Hcc2μP+Htc2μS)(xi,vi)iIP(Hct11μP+Htt11μS)(xi,vi)+(Hct2μP+Htt2μS)(xi,vi)iIS. (2.9)

    Inspired by [33], we design a lane-changing rule based on a trade-off between the expected own advantage and the disadvantage imposed on other drivers. In particular, the follower in the target lane is involved in the decision process. The subjective utility of a change of lane increases with the gap to the new leader in the target lane. However, if the velocity of this leader is lower, it may be convenient to stay in the present lane. A criterion for the utility including the above mentioned situations is the difference in the accelerations after and before the lane change. The formulation in terms of accelerations has several advantages. Indeed the evaluation of the traffic situation is transferred to the acceleration function of the Bando-FtL model with the result that the lane change rules are compact and depend only on a small number of additional parameters.

    Now we consider P cars and S trucks on an open stretch road of LN>0 lanes. Let K={1,,L}. First of all, we consider the lane-changing condition in a homogeneous traffic flow. Let Δ>0 be fixed. The choice of Δ depends on the vehicle type in the homogeneous traffic flow. Let aki(t) be the acceleration of vehicle i on the current lane kK at time t[0,T], and ˉaki(t) the expected acceleration of vehicle i on the adjacent lane k=k+1 or k1 at time t. In addition, We assume that the accelerations and expected accelerations (if there is lane-changing) of vehicles in the homogeneous traffic flow are bounded from above by MRΔ. Denote ikL, ikF the label of the leading and following vehicle of i-th vehicle on the adjacent lane k, respectively, if vehicle i performs lane-changing from lane k to lane k and ˉakikF(t) the expected acceleration of vehicle ikF at time t[0,T].

    In a homogeneous traffic flow, vehicle i will perform lane changing at time t[0,T] from lane k to lane k with probability p([ˉaki(t)aki(t)Δ]+,[ˉaki(t)+Δ]+,[ˉakikF(t)+Δ]+) if the following conditions are satisfied

    Incentive: ˉaki(t)aki(t)+ΔSafety: ˉaki(t)Δ and ˉakikF(t)Δ.

    Here Δ represents a lower bound for the difference of the expected acceleration and current acceleration which determines the incentive of a vehicle to perform a lane-change. In particular, if the expected acceleration of vehicle i on its neighbor lane k is at least Δ bigger than its actual acceleration on its current lane k, then vehicle i has the incentive to perform lane-changing from lane k to lane k. The safety condition guarantees that there is no excessive breaking for both vehicle i and its new follower ikF on the adjacient lane k if vehicle i changing from lane k to lane k.

    Furthermore, a possible choice of the probability function is

    p:(R+)3[0,1] with p(b1,b2,b3)=1C(1eγb1b2b3)[0,1],γ>0, (2.10)

    where C is a renormalization constant defined as

    C=max[0,2M+Δ]3(1eγb1b2b3)=1eγ(2M+Δ)3.

    But our result will be valid for any function having the following properties:

    ● It strictly increases with respect to each one of its input;

    ● If one of its input is zero, then the probability function value is zero.

    Now we will consider the lane-changing condition in a heterogeneous traffic flow encompassing cars and trucks. Specifically, we need to modify the "incentive" and "safety" conditions according to the different car-following combinations and vehicle types.

    Let Δn>0, with n{cc,ct,tc,tt,c,t}, be fixed. Vehicle iI will change from lane k to lane k at time t[0,T] with a given certain probability (to be defined later) if the following conditions are satisfied:

    Incentive: ˉaki(t){aki(t)+Δcc if i,ikFIP,aki(t)+Δtc if iIP,ikFIS,aki(t)+Δct if iIS,ikFIP,aki(t)+Δtt if i,ikFIS; (2.11)
    Safety: ˉaki(t){Δc and ˉakikF(t)Δc if i,ikFIPΔc and ˉakikF(t)Δt if iIP,ikFIS,Δt and ˉakikF(t)Δc if iIS,ikFIP,Δt and ˉakikF(t)Δt if i,ikFIS. (2.12)

    For instance, the probability of vehicle iIP performing lane-changing from lane k to lane k with its new follower on lane k being a truck, i.e., ikFIS, at time t[0,T] is

    p([ˉaki(t)aki(t)Δtc]+,[ˉaki(t)+Δc]+,[ˉakikF(t)+Δt]+)

    and the probability of vehicle iIS performing lane-changing from lane k to lane k with its new follower on lane k being a car, i.e., ikFIP, at time t[0,T] is

    p([ˉaki(t)aki(t)Δct]+,[ˉaki(t)+Δt]+,[ˉakikF(t)+Δc]+).

    Here the probability function p is defined in Eq (2.10) with renormalization constant

    C=max[0,2M+Δ]3(1eγb1b2b3)=1eγ(2M+Δ)3,

    where Δ=min{Δcc,Δct,Δtc,Δtt,Δc,Δt} and MRΔ is a common upper bound for the acceleration of both cars and trucks, i.e., for every t[0,T] and iI, |ai(t)|<M.

    Note that in the heterogeneous traffic condition, each acceleration aki, ˉaki, and ˉakikL has four different alternatives based on the four different car-truck car-following combinations. Furthermore, by Eq (2.3), each acceleration aki, ˉaki, and ˉakikL depends on the space gap, the velocity of the reference vehicle and the velocity of the leader of the reference vehicle. The incentive condition defined in Eq (2.11) implies that, before changing lane, vehicle i needs to check its space gap, velocity and velocity difference with its leading vehicle on the current and adjacent lane. The safety condition defined in Eq (2.11) implies that there is no excessive breaking for vehicle i and its follower ijL on the adjacent lane k after lane-changing.

    In this subsection, we recall some notions and properties related to the generalized Wasserstein distance [38] and give the statement of a revised version of Ascoli-Arzelá theorem [23].

    The Generalized Wasserstein Distance.

    In the following we denote with

    d the dimension of the space;

    M the space of Borel measures with finite mass on Rd;

    suppμ the support of measure μM;

    P the space of probability measures (the measures in M with unit mass) on Rd;

    Mp the space of Borel measures with finite p-th moment on Rd;

    Mac0M the subset of measures that are absolutely continuous with respect to the Lebesgue measure with bounded support.

    Given a measure μM, we denote its mass by |μ|:=μ(Rd). Given a Borel map γ:RdRd, the push-forward of μM by γ, γ#μ, is defined as for every Borel set ARd, γ#μ(A):=μ(γ1(A)). One can see that the mass of γ#μ is identical to the mass of μ, i.e., |μ|=|γ#μ|.

    We use the notation μ1μ when μ1 is absolutely continuous with respect to μM and for every Borel set ARd, μ1(A)μ(A).

    Now we recall the definition of the generalized Wasserstein distance on M.

    Definition 2.1. Given a,b(0,) and p1, the generalized Wasserstein distance between two measures μ,νMp is

    Wa,bp(μ,ν):=inf˜μ,˜νMp|˜μ|=|˜ν|(a(|μ˜μ|+|ν˜ν|)+bWp(˜μ,˜ν)), (2.13)

    where Wp(˜μ,˜ν) is the Wasserstein distance between the measures ˜μ,˜νMp with |˜μ|=|˜ν|.

    We recall that the standard Wasserstein distance is defined only for Borel measures with the same mass and combining it with the L1 distance we get the generalized Wasserstein distance which can be applied instead to measures with different masses.

    Remark 2.1. Note that the infimum on the right hand side of Eq (2.13) is actually a minimum if one takes ˜μμ and ˜νν.

    We recall the following key result (Proposition 2 in [38]).

    Proposition 2.2. The following properties of the generalized Wasserstein distance W1,11 hold for measures μ,ν,μ1,μ2,ν1,ν2Mp

    W1,11(kμ,kν)kW1,11(μ,ν)fork0;W1,11(μ1+μ2,ν1+ν2)W1,11(μ1,ν1)+W1,11(μ2,ν2).

    To be self-contained, we list the following two lemmata from [23].

    Lemma 2.3. For every f,g:RnRn bounded Borel measureable functions and μM1(Rn), one has

    W1,11(f#μ,g#μ)fgL(suppμ).

    Moreover if f is a locally Lipschitz continuous Borel measurable function with Lipschitz constant L on the ball B of Rn, then for μ,νM1(Rn) compactly supported on B,

    W1,11(f#μ,f#ν)max{L,1}W1,11(μ,ν).

    Lemma 2.4. Let H be a locally Lipschitz continuous map with sub-linear growth. Let R>0 be fixed, and d be the dimension of the space, and μ,ν:[0,T]M1(Rd) be continuous maps with respect to the generalized Wasserstein distance W1,11 such that for every t[0,T]

    suppμ(t)B(0,R)andsuppν(t)B(0,R).

    For every ρ>0, there exists a constant Lρ,R such that

    Hμ(t)Hν(t)L(B(0,ρ))Lρ,RW1,11(μ(t),ν(t)). (2.14)

    An extended version of Ascoli-Arzelá theorem

    In the following we recall an extended version of Ascoli-Arzelá theorem from [23].

    Theorem 2.5. Let K be a compact subset of R and let E be a complete and totally bounded metric space with metric d. Consider a sequence of functions {fn}n=1 in C(K;E). If there exists L>0, such that the following is true: for any ε>0, there exists N>0, such that, whenever nN,

    d(fn(t),fn(s))L|ts|+min{ε,|ts|},s,tK

    then the sequence {fn}n=1 has a uniformly convergent sub-sequence.

    In this subsection, we start with a finite dimensional ODE system and derive its mean-field limit formally.

    Let R>0 be fixed. Let M,NN+. Denote with D the domain R×R0, with DM=RM×RM0 and DN=RN×RN0. Let Hi:DR with i=1,,8 be locally Lipschitz continuous maps with sub-linear growth. Then given an initial datum I0:=(x0,v0,y0,w0)DM×DN, there exists a unique solution I(t):=(x(t),v(t),y(t),w(t))DM×DN on the whole time interval [0,T] to the following system of ODEs on DM×DN

    ˙xi(t)=vi(t),i=1,,M˙vi(t)=(H11μM+H21νN+H3μM+H4νN)(xi,vi)˙yj(t)=wj(t),j=1,,N˙wj(t)=(H51μM+H61νN+H7μM+H8νN)(yj,wj) (2.15)

    where μM,νN:[0,T]P(D)M1(D) are defined as follows

    μM(t)=1MMi=1δ(xi(t),vi(t)),νN(t)=1NNj=1δ(yj(t),wj(t)). (2.16)

    For more details, we refer the readers to [18].

    Let us further assume that for each t[0,T], the empirical measures μM(t),νN(t) in P(D)M1(D) are with uniform support in both M and N. By Prohorov's theorem (see [6]) it follows that the sequences (μM)M and (νN)N are weakly* relatively compact. Therefore, there exist subsequences (μMk)k, (νNk)k and μ,νP(D)M1(D) such that

    μMkμ as kνNkν as k (2.17)

    with weak* convergence in P(D)M1(D) pointwise in time.

    Now we take M,N in Eq (2.15) and derive the mean-field limit of the finite-dimensional ODE system formally. Let us consider a test function φC10(D) and we compute

    ddtμM(t),φ=1MMi=1ddtφ(xi(t),vi(t))=1MMi=1(xφ(xi(t),vi(t))vi(t)+vφ(xi(t),vi(t))˙vi(t))=1MMi=1xφ(xi(t),vi(t))vi(t)++1MMi=1(vφ(xi(t),vi(t))(H11μM+H21νN+H3μM+H4νN)(xi,vi))=μM(t),xφ(x,v)v+μM(t),vφ(x,v)(H11μM+H21νN+H3μM+H4νN)(x,v)=xμM(t)v,φv(H11μM+H21νN+H3μM+H4νN)(x,v)μM(t),φ

    which implies

    tμM(t)+vxμM(t)+v[(H11μM+H21νN+H3μM+H4νN)(x,v)μM(t)]=0. (2.18)

    In addition, we have

    ddtνN(t),φ=1NNj=1ddtφ(yj(t),wj(t))=1N(Nj=1yφ(yj(t),wj(t))wj(t)+Nj=1wφ(yi(t),wi(t))˙wj(t))=1NNj=1yφ(yj(t),wj(t))++1NNj=1(wφ(yj(t),wj(t))(H51μM+H61νN+H7μM+H8νN)(yj,wj))=νN(t),yφ(y,w)w+νN(t),wφ(y,w)(H51μM+H61νN+H7μM+H8νN)(y,w)=yνN(t)w,φ(y,w)w(νN(t)(H51νN+H61νN+7μM+H8νN)(y,w)),φ

    from which we conclude

    tνN(t)+wyνN(t)+w[(H51μM+H61νN+H7μM+H8νN)(y,w)νN(t)]=0. (2.19)

    Combine with Eqs (2.17), (2.18) and (2.19), for the limit of k of the subsequences (μMk)k and (νNk)k, formally we have

    tμ(t)+vxμ(t)+v[(H11μ+H21ν+H3μ+H4ν)(x,v)μ(t)]=0,
    tν(t)+wyν(t)+w[(H51μ+H61ν+H7μ+H8ν)(y,w)ν(t)]=0.

    In the last section, we recall some related results on partial differential equations of Vlasov-type with and without source terms.

    A family of Lipschitz continuous flow maps is associated to the system (2.15)

    Tμ,νt:I0DM×DNI(t)DM×DN. (2.20)

    indexed by t[0,T]. For more details we refer to [18].

    Given the initial conditions (μ0,ν0)(P(D)M1(D))2 with bounded support, we say that the couple of measures (μ(t),ν(t)) is a weak equi-compactly supported solution of the following Vlasov-type PDE system with the initial datum (μ0,ν0),

    tμ+vxμ+v[(H11μ+H21ν+H3μ+H4ν)μ]=0,tν+vxν+v[(H51μ+H61ν+H7μ+H8ν)ν]=0, (2.21)

    if (i)μ(0)=μ0 and ν(0)=ν0;

    (ii)suppμ(t),suppν(t)BD(0,R) for all t[0,T];

    (iii)for every φCc(R2),

    ddtDφ(x,v)dμ(t)(x,v)=Dφ(x,v)˜ω(t,x,v)dμ(t)(x,v),
    ddtDφ(y,w)dν(t)(y,w)=Dφ(y,w)ˆω(t,y,w)dν(t)(y,w)

    where ˜ω(t,x,v)=˜ωH1,H2,H3,H4,μ,ν(t,x,v):[0,T]×DR2 is defined as

    ˜ωH1,H2,H3,H4,μ,ν(t,x,v):=(v,(H11μ+H21ν+H3μ+H4ν)(x,v)), (2.22)

    and ˆω(t,y,w)=ˆωH5,H6,H7,H8,μ,ν(t,y,w):[0,T]×DR2 is defined as

    ˆωH5,H6,H7,H8,μ,ν(t,y,w):=(v,(H51μ+H61ν+H7μ+H8ν)(y,w)). (2.23)

    Furthermore, following from Section 8.1 in [2], the couple of measures (μ(t),ν(t)) is a weak equi-compactly supported solution of the system (2.21) if and only if it satisfies condition (ii) and the measure-theoretical fixed point equation (μ(t),ν(t))=(Tμ,νt)#(μ0,ν0) where the flow function Tμ,νt is defined in Eq (2.20).

    Now we consider solutions to the following Vlasov-type PDE system with initial datum (μ0,ν0)(Mac0(D)M1(D))2, and source terms G1 and G2

    tμ+vxμ+v[(H11μ+H21ν+H3μ+H4ν)μ]=G1(μ,ν)tν+vxν+y[(H51μ+H61ν+H7μ+H8ν)ν]=G2(μ,ν) (2.24)

    under the following hypotheses:

    (A1)G1(μ,ν),G2(μ,ν) have uniformly bounded mass and support, that is, there exist Q,R, such that |G1(μ,ν)|(D),|G2(μ,ν)|(D)Q, and supp(G1(μ,ν)),supp(G2(μ,ν))BD(0,R);(A2)G1 and G2 are Lipschitz, that is, there exists L, such that, for any μ,μ,ν,νM1(D),W1,11(Gi(μ,ν),Gi(μ,ν))L(W1,11(μ,μ)+W1,11(ν,ν)),i=1,2.

    A coupled of measures (μ(t),ν(t)) are weak solutions of Eq (2.24) with a given initial datum (μ0,ν0)(Mac0(D)M1(D))2, if μ(0)=μ0,ν(0)=ν0 and if for every φCc(R2), it holds

    ddtDφ(x,v)dμ(t)(x,v)==Dφ(x,v)dG1(μ,ν)(x,v)+D1φ(x,v)˜ωH1,H2,H3,H4,μ,ν(t,x,v)dμ(t)(x,v),
    ddtDφ(y,w)dμ(t)(y,w)==Dφ2(y,w)dG2(μ,ν)(y,w)+Dφ(y,w)ˆωH5,H6,H7,H8,μ,ν(t,y,w)dν(t)(y,w),

    where ˜ω,ˆω is as defined in Eqs (2.22) and (2.23). We have the following:

    Theorem 2.6. Given an initial datum (μ0,ν0)(Mac0(D)M1(D))2, under the hypotheses (A1) and (A2), there exists a unique weak solution (μ(t),ν(t)) to the system (2.24) in (Mac0(D)M1(D))2.

    A weak solution (μ(t),ν(t)) to Eq (2.24) can be constructed using a sample-and-hold Lagrangian scheme. For a fixed jN+, define Δt:=T2j and decompose the time interval [0,T] in [0,Δt),[Δt,2Δt),,[(2j1)Δt,2jΔt), define

    Initial step (μj(0),νj(0)):=(μ0,ν0);

    Recursive step 1 (μj((n+1)Δt),νj((n+1)Δt)):=Tμj(nΔt),ν(nΔt)Δt#(μj(nΔt),

    νj(nΔt)) +Δt(G1(μj(nΔt),νj(nΔt)),G2(μj(nΔt),νj(nΔt)));

    Recursive step 2 (μj(t),νj(t)):=Tμj(nΔt),ν(nΔt)τ#(μj(nΔt),νj(nΔt))+

    +τ(G1(μj(nΔt),νj(nΔt)),G2(μj(nΔt),νj(nΔt)));

    where n is the maximum integer such that tnΔt0 and τ:=tnΔt. Then (μ(t),ν(t))=limj(μj(t),νj(t)) is the unique weak solution to Eq (2.24). For more details, please see [38].

    In order to describe the frequencies of the vehicles' lane change behavior and prove the well-posedness of our heterogeneus multi-lane traffic model, it is critical to introduce the model assumption "cool-down" time. Indeed, empirical observations showed that the lane-changing frequency of vehicles on the highway is low. A key example is a study done on the German highways which shows that only 15% of the vehicles performes lane-change while traveling the recorded road segment [29]. For this reason, the chance of two vehicles performing lane-change at exactly the same time is even lower and it is reasonable to assume that this does not happen at all. In the next we state mathematically what just explained.

    Each vehicle iI is associated to a timer τi and the initial timer differs from vehicle to vehicle. We introduce a "cool-down" time ˉτ=TNτ, with NτN+ large. Every vehicle checks the lane-changing conditions only when its timer reaches ˉτ. When this happens, the vehicle's timer is then set to 0. More explicitly, for each vehicle iI, its timer τi satisfies the differential equation

    ˙τi(t)=1,τi(0)=τi,0,t[0,ˉτ)

    with the following assumption on the initial data:

    ijIτi,0τj,0. (3.1)

    When t=ˉτ we set τi(t)=0. We can also model a large lane-change frequency by simply choosing a small cool-down time ˉτ.

    In the case of finitely many vehicles, the presence of the cool-down time, ˉτ allows us to consider a small time interval [0,t1] during which there is no vehicle changing lane, with

    t1=miniI{ˉττi,0}

    Consider the space X=R×R0×[0,ˉτ) and the set L={=(i)iIKP+S} of symbols that represent all possible lane labels of all vehicles among cars and trucks.

    Let AX be the set of triples position-velocity-timer of all vehicles among which there are at least two vehicles occupying the same lane and position at certain time, i.e.,

    A={(xi,vi,τi)iIX:t[0,T],i1,i2I,s.t.,xi1(t)=xi2(t)i1(t)=i2(t), with i1,i2K}, (3.2)

    As in Section 2, let I={1,,P+S} be the set of index for all the vehicles, IP={1,,P} the set of index for cars and IS={P+1,,P+S} for trucks. Denote respectively with with IkP(t) and IkS(t) the set of indices for cars and trucks on lane k at time t[0,T], with Pk(t) and Sk(t) the number of cars and trucks on lane k at time t[0,T].

    The time dependent atomic probability measures on the k lane are given by

    μkP(t)=1Pk(t)iIkP(t)δ(xi(t),vi(t)),μkS(t)=1Sk(t)iIkS(t)δ(xi(t),vi(t)). (3.3)

    where (xi(t),vi(t)) are solutions of the following first order system:

    ˙xi=viiI,˙vi={(Hcc11μkP+Htc11μkS)(xi,vi)+(Hcc2μkP+Htc2μkS)(xi,vi)iIkP(Hct11μkP+Htt11μkS)(xi,vi)+(Hct2μkP+Htt2μkS)(xi,vi)iIkS. (3.4)

    And finally consider the switching set LC(Σ) describing the lane-changing mechanism of the finitely many vehicles:

    LC(Σ)={(,(xi,vi,τi),,(xi,vi,τi))iI(L×X)2:i0I,t0[0,ˉτ),s.t.,ji0,(j(t0),xj(t0),vj(t0),τj(t0))=(j(t0),xj(t0),vj(t0),τj(t0))(xi0(t0),vi0(t0))=(xi0(t0),vi0(t0)),i0(t0)=i0(t0)±1,τi0(t0)=0}. (3.5)

    Now we are ready to give the definition of hybrid system.

    Definition 3.1. A hybrid system is a 4-tuple Σ=(L,M,g,SW) where:

    (1) L={=(i)iIKP+S} is a finite set of symbols that represent all possible lane labels of all vehicles;

    (2) M={M}L, where M=(XA)P+S, with A defined in Eq (3.2).

    (3) g={g}L, g:MR3(P+S), g(i)=(vi,ai,1), where ai=˙vi as defined in systems (3.4);

    (4) SWLC(Σ) is the set of states for which a lane-changing actually occurs.

    We need two further definitions before stating and proving the result of existence of solutions for the hybrid system.

    Definition 3.2. A hybrid state of the hybrid system Σ is a 4-tuple (,x,v,τ)L×M. The set of all the hybrid states of the hybrid system Σ will be called HS.

    Definition 3.3. Let (0,x0,v0,τ0)(K×X)P+S be an initial condition to the hybrid system Σ and assume that τ0 satisfies Eq (3.1). A trajectory of the hybrid system Σ with initial condition (0,x0,v0,τ0) is a map φ:[0,T]HS, φ(t)=((t),x(t),v(t),τ(t)), such that

    (1) ((0),x(0),v(0),τ(0))=(0,x0,v0,τ0);

    (2) i[0,ˉττi,0)=i,0,iI;

    i[nˉττi,0,(n+1)ˉττi,0)=i,nL,iI;

    (3) τi(nˉττi,0)=0iI;

    (4) limt(nˉττi,0)xi(t)=xi(nˉττi,0);

    (5) For almost every t[0,T]

    ddt(xi,vi,τi)=gi(t)(xi(t),vi(t),τi(t))iI. (3.6)

    Theorem 3.1 (Existence and uniqueness of trajectories to the hybrid system Σ). Let Hn1:R×R0R, and Hn2:R×R0R with n{cc,ct,tc,tt} be locally Lipschitz convolution kernels with sub-linear growth and let (0,x0,v0,τ0)(K×X)P+S be a given initial datum. Then there exists a unique trajectory φ:[0,T]HS to hybrid system Σ, which is also Lipschitz continuous in time over the time interval in which no lane change occurs.

    Proof. Let t0=miniI{ˉττi,0} where τi,0 is the i-th component of the vector τ0[0,ˉτ)P+S. By definition no lane change is performed in the time interval [0,t0) and the dynamic of each vehicle in the lane k is given by Eq (3.4). More compactly, we can call φk(t)=(xk(t),vk(t))(R×R0)Pk+Sk the trajectory of vehicles on the lane k over the time interval [0,t0) and re-write the system (3.4) as

    ˙φk(t)=gk(t,φk(t)). (3.7)

    Here gk:[0,t0)×(R×R0)Pk+Sk(R×R0)Pk+Sk is defined as

    gk(t,φk(t))=(vk(t),ak(t)),

    with

    aki(t)={(Hcc11μkP+Htc11μkS)(xi,vi)+(Hcc2μkP+Htc2μkS)(xi,vi)iIkP(Hct11μkP+Htt11μkS)(xi,vi)+(Hct2μkP+Htt2μkS)(xi,vi)iIkS.

    By the regularity and growth assumptions on the convolution kernels, it is immediate to check that

    gk(t,φk(t))C(1+φk(t)) (3.8)

    for a constant C which does not depend on the number of vehicles (see Lemma 3.4 in [23] for details). Therefore the Caratheodory Theorem [24] yields the existence of solution φk to the linear system (3.7) on the time interval [0,t0) with initial data φk0=(xk0,vk0)(R×R0)Pk+Sk. Moreover the solution satisfies the following growth condition

    φk(t)(φk0+Ct0)eCt0 (3.9)

    which implies also the Lipschitz continuity. Indeed for any times t,t[0,t0) we have

    φk(t)φk(t)ttgk(s,φk(s))dsttC(1+φk(s))ds|C(1+(φk0+Ct0)eCt0)|tt|.

    Analogously, on all the finitely time intervals in which there is no lane change, the Caratheodory Theorem still yields the existence of a unique Lipschitz trajectory for vehicles in the same lane.

    In the next, we let the number of cars and trucks approach infinity. The emerging equations do not describe anymore the trajectories of the single vehicle but the evolution of density of each class of vehicles in space and velocity.

    For convenience, we introduce the following compact notation for Eq (2.9):

    ˙xi=vi˙vi=((Hn111μP+Hn211μS+Hn12μP+Hn22μS)(xi,vi) (4.1)

    with (n1,n2)=(cc,tc) if iIP and (n1,n2)=(ct,tt) if iIS.

    What we are going to prove is that the mean field limit of the hybrid system in Definition 3.1 is a system of two Vlasov-type equations with source terms. These source terms are generated by the lane-change behaviour in the four different car-truck car-following combinations and induce the measure solutions to change mass in time, therefore the limit is obtained by using the generalized Wasserstein distance. In detail we will derive the following limit system

    tνkc+vxνkc+v[(Hcc11νkc+Htc11νkt+Hcc2νkc+Htc2νkt)νkc]=G1(νkc,νkt,νkc,νkt), (4.2)
    tνkt+vxνkt+v[(Hct11νkc+Htt11νkt+Hct2νkc+Htt2νkt)νkt]=G2(νkc,νkt,νkc,νkt), (4.3)

    where νkc and νkt represent respectively the density of cars and trucks on the lane k. To describe the derivation process of the source terms G1 and G2, we need to introduce the average accelerations AkP and AkS defined as

    AkP=Hcc11μkP+Htc11μkS+Hcc2μkP+Htc2μkS,
    AkS=Hct11μkP+Htt11μkS+Hct2μkP+Htt2μkS,

    with μkP and μkS the probability measures given in Eq (2.4) on lane k. Since μkP, μkS are both compactly supported and the convolution kernels are, by assumption, locally Lipschitz and with sub-linear growth, it follows that both the average accelerations are bounded.

    We define the map p1 as

    p1(b1,b2,b3,b4,b5)=1C(1eγ2b1b2b3b4b5),
    p1([AkPAkPΔcc]+,[AkSAkPΔtc]+,[AkP+Δc]+,[AkS+Δt]+,[AkS+Δt]+),

    representing the probability of cars performing lane change from lane k to lane k and analogously the map p2 as

    p2(b1,b2,b3,b4,b5)=1C(1eγ2b1b2b3b4b5)
    p2([AkPAkSΔct]+,[AkSAkSΔtt]+,[AkP+Δc]+,[AkS+Δt]+,[AkS+Δt]+)

    which is instead the probability of trucks performing lane change from lane k to lane k.

    Looking closely at the probability p1 (the same considerations also apply to p2), we can observe that it is strictly positive only if the safety and incentive conditions are strictly satisfied in an average sense, i.e., if

    AkP>AkP+Δcc,AkS>AkP+Δtc,AkP>Δc,and AkS>Δt.

    Thanks to this notation we can define the source terms as

    G1(νkc,νkt,νkc,νkt)=[Gk1,k1(νk1c,νk1t,νkc,νkt)Gk,k11(νk1c,νk1t,νkc,νkt)](1δ1(k))+[Gk+1,k1(νk+1c,νk+1t,νkc,νkt)Gk,k+11(νk+1c,νk+1t,νkc,νkt)](1δN(k))
    G2(νkc,νkt,νkc,νkt)=[Gk1,k2(νk1c,νk1t,νkc,νkt)Gk,k12(νk1c,νk1t,νkc,νkt)](1δ1(k))+[Gk+1,k2(νk+1c,νk+1t,νkc,νkt)Gk,k+12(νk+1c,νk+1t,νkc,νkt)](1δN(k))

    where k{k1,k+1} and Gj,j1, Gj,j2 are respectively given by

    Gj,j1(νjc,νjt,νjc,νjt)=p1([AjPAjPΔcc]+,[AjSAjPΔtc]+,[AjP+Δc]+,[AjP+Δc]+,[AjS+Δt]+)νjc,Gj,j2(νjc,νjt,νjc,νjt)=p2([AjPAjSΔct]+,[AjSAjSΔtt]+,[AjS+Δt]+,[AjP+Δc]+,[AjS+Δt]+)νjt,

    with j,j{k1,k,k+1}.

    Definition 4.1 (Weak solution to the coupled PDEs). Given initial datum (νc,0,νt,0)(Mac0(D)M1(D))2L, we say that (νc,νt):[0,T](Mac0(D)M1(D))2L is a solution to the coupled PDEs (4.2), (4.3), if for every test function φCc(D) and for all k{1,,L}, νkc and νkt are compactly supported in B(0,R) for some R>0, and for almost every t[0,T],

    ddtR×R0φ(x,v)dνkc(t)(x,v)==R×R0φ(x,v)dG1(νkc,νkt,νkc,νkt)(t)(x,v)+R×R0(φ(x,v)ωkc(t,x,v))dνkc(t)(x,v) (4.4)
    ddtR×R0φ(x,v)dνkt(t)(x,v)==R×R0φ(x,v)dG2(νkc,νkt,νkc,νkt)(t)(x,v)+R×R0(φ(x,v)ωkt(t,x,v))dνkt(t)(x,v) (4.5)

    where

    ωkc(t,x,v)=(v,(Hcc11νkc+Htc11νkt+Hcc2νkc+Htc2νkt)(x,v)),

    and

    ωkt(t,x,v)=(v,(Hct11νkc+Htt11νkt+Hct2νkc+Htt2νkt)(x,v)).

    Theorem 4.1 (Existence of weak solutions to the couple PDEs). Let the initial datum (νc,0,νt,0)(Mac0(D)M1(D))2L be given. Assume that convolutional kernels Hnq with q=1,2 and n{cc,ct,tc,tt} are locally Lipschitz and with sub-linear growth. Then there exists a solution (νc,νt):[0.T](Mac0(D)M1(D))2N to the coupled PDEs (4.2), (4.3) as in Definition 4.1.

    Proof. As first step we construct a sequence of discrete measures converging to the initial datum in the generalized Wasserstein distance. Indeed on each lane k{1,,L}, there exists a infinite set of couples (xki,0,vki,0)R×R0, such that

    νkc,0=limpkmciIkPδ(xki,0,vki,0) (4.6)

    and

    νkt,0=limskmtiIkSδ(xki,0,vki,0), (4.7)

    where νc,0=(νkc,0)Lk=1 and νt,0=(νkt,0)Lk=1, IkP is a finite subset of indices for cars and IkS is a finite subset of indices for trucks on lane k, while pk=#IkP and sk=#IkS represent the number of cars and trucks on lane k respectively. The constants mc and mt are the average masses for cars and trucks defined as

    mc=Lj=1νjc,0Lj=1pj,mt=Lj=1νjt,0Lj=1sj.

    Let P be the set of cars and S the set of trucks on the open stretch road. The couples (x0,v0)=(xi,0,vi,0), with iI (set of indices for both vehicles) introduced for the approximation represent the initial positions and velocities of cars and trucks. We also define the following multi-valued functions:

    IkP():[0,T]P(I) and IkS():[0,T]P(I)

    where P() stands for power set. These keep trace of the set of the indices for cars and trucks on each lane on the time interval [0,T]. To each one of these multi-functions it is naturally associated a map counting the number of cars and number of trucks present on the lane k at any time, i.e.,

    Pk():[0,T]N+,Pk(t)=#IkP(t),
    Sk():[0,T]N+,Sk(t)=#IkS(t).

    with Pk(0)=pk and Sk(0)=sk. By Theorem 3.1, we know that for t[0,T] and iIi(t)P(t), there exists a unique map (xi,vi):[0,T]proj1,2(Mi(t)) (where proj1,2:R3R2;(x,y,z)(x,y)) which represents the positions and velocities of cars on lane i(t) during the time interval [0,T] with (xi(0),vi(0))=(xi,0,vi,0). Define a discrete measure

    νpk(t)=mciIkP(t)δ(xi(t),vi(t)). (4.8)

    Similarly, for t[0,T], iIi(t)S(t), there exists a unique map (xi,vi):[0,T]proj1,2(Mi(t)) representing positions and velocities of trucks on lane i(t) during the time interval [0,T] with (xi(0),vi(0))=(xi,0,vi,0). We can define a discrete measure

    νsk(t)=mtiIkS(t)δ(xi(t),vi(t)). (4.9)

    Note that mc0 as pk. Therefore there exists a constant L>0 which satisfies the following condition: for every ε>0, we can find N1>0, such that whenever pk>N1,

    W1,11(νpk(s),νpk(t))<L|st|+min{ε,|st|}s,t[0,T].

    By Theorem 2.5, there exist a convergent subsequence (νpk) (for simplicity, we use the some notation for the subsequence as the notation for the original sequence) and νkcM(D) such that

    νpkνkc as pk.

    Analogously we can conclude that there exists a subsequence of νsk converging to νktM(D) as sk. Next we will show that (νkc,νkt)(M(D))2 with k{1,,N}, is a weak solution to the coupled PDEs (4.2) and (4.3) as in Definition 4.1. Let IkP1 be the set of indices of cars on lane k not performing lane-change over the whole time interval [0,T] and set pk1=|IkP1|. Consider the following discrete measure to track positions and velocities for cars in this set:

    νpk1(t)=mciIkP1δ(xi(t),vi(t)).

    Then, for any test function φC(D) we have

    ddtφ,νpk1=ddtmciIkP1φ(xi(t),vi(t))=mciIkP1(xφ(xi(t),vi(t))vi(t)+vφ(xi(t),vi(t))˙vi(t))=mciIkP1xφ(xi(t),vi(t))vi(t)+mciIkP1vφ(xi(t),vi(t))(Hcc11νpk(t)+Htc11νsk(t)+Hcc2νpk(t)+Htc2νsk(t))(xi(t),vi(t))=xφ(x,v)v,νpk1+vφ(x,v)(Hcc11νpk(t)+Htc11νsk(t)+Hcc2νpk(t)+Htc2νsk(t))(x,v),νpk1. (4.10)

    For all t[0,T] and s[0,t], by integrating both sides of Eq (4.10) we get

    φ,νpk1(s)νpk1(0)=s0[R×R+xφ(x,v)v+vφ(x,v)(Hcc11νpk(t)+Htc11νsk(t)+Hcc2νpk(t)+Htc2νsk(t))(x,v)dνpk1(t)(x,v)]dt. (4.11)

    Analogously, let IkS1 be the set of indices of trucks on lane k not performing lane-change over the whole time interval [0,T] and set sk1=|IkS1|. Consider the following discrete measure which keeps trace of positions and velocities for these trucks:

    νsk1(t)=mtiIkS1δ(xi(t),vi(t)).

    For any test function φC(R×R0) we have

    ddtφ,νsk1=ddtmtiIkS1φ(xi(t),vi(t))=mtiIkS1(xφ(xi(t),vi(t))vi(t)+vφ(xi(t),vi(t))˙vi(t))=mtiIkS1xφ(xi(t),vi(t))vi(t)+mtiIkS1vφ(xi(t),vi(t))(Hct11νpk(t)+Htt11νsk(t)+Hct2νpk(t)+Htt2νsk(t))(xi(t),vi(t))=xφ(x,v)v,νsk1+vφ(x,v)(Hct11νpk(t)+Htt11νsk(t)+Hct2νpk(t)+Htt2νsk(t))(x,v),νsk1 (4.12)

    For all t[0,T] and s[0,t], by integrating both sides of Eq (4.12)

    φ,νsk1(s)νsk1(0)=s0[R×R+xφ(x,v)v+vφ(x,v)(Hct11νpk(t)+Htt11νsk(t)+Hct2νpk(t)+Htt2νsk(t))(x,v)dνsk1(t)(x,v)]dt. (4.13)

    Let pk1 and sk1 (respectively the number of cars and trucks not performing lane change on lane k) go to infinity, then on the left hand side of Eqs (4.11)–(4.13) we have

    limpk1φ,νpk1(s)νpk1(0)=φ,νkcνkc,0, (4.14)
    limsk1φ,νsk1(s)νs1k(0)=φ,νktνkt,0. (4.15)

    By the dominated convergence theorem, on the right hand side of Eqs (4.11)–(4.13) we have that for all test function φCc(D),

    limpk1s0(R×R+xφ(x,v)v)dνpk1(t)(x,v)dt=s0(R×R+xφ(x,v)v)dνkc(t)(x,v)dt (4.16)
    limsk1s0(R×R+xφ(x,v)v)dνsk1(t)(x,v)dt=s0(R×R+xφ(x,v)v)dνkt(t)(x,v)dt (4.17)

    and

    limsk1limpk1s0(vφ(x,v)(Hcc11νpk(t)+Htc11νsk(t)+Hcc2νpk(t)+Htc2νsk(t))(x,v))dνpk1(t)(x,v)dt=s0(vφ(x,v)(Hcc11νkc(t)+Htc11νkt(t)+Hcc2νkc(t)+Htc2νkt(t))(x,v))dνkc(t)(x,v)dt, (4.18)
    limsk1limpk1s0(vφ(x,v)(Hct11νpk(t)+Htt11νsk(t)+Hct2νpk(t)+Htt2νsk(t))(x,v))dνsk1(t)(x,v)dt=s0(vφ(x,v)(Hct11νkc(t)+Htt11νkt(t)+Hct2νkc(t)+Htt2νkt(t))(x,v))dνkt(t)(x,v)dt. (4.19)

    Indeed, for every r>0, Lemma 2.4 yields

    limsk1limpk1(Hcc11νpk(t)+Htc11νsk(t)+Hcc2νpk(t)+Htc2νsk(t))(x,v)(Hcc11νkc(t)+Htc11νkt(t)+Hcc2νkc(t)+Htc2νkt(t))(x,v)L(B(0,r))=0, (4.20)
    limsk1limpk1(Hcc11νpk(t)+Hct11νsk(t)+Hct2νpk(t)+Htt2νsk(t))(x,v)(Hct11νkc(t)+Htt11νkt(t)+Hct2νkc(t)+Htt2νkt(t))(x,v)L(B(0,r))=0. (4.21)

    Since φCc(R×R0),

    limsk1limpk1vφ(x,v)[(Hcc11νpk(t)+Htc11νsk(t)+Hcc2νpk(t)+Htc2νsk(t))(x,v)(Hcc11νkP(t)+Htc11νkt(t)+Hcc2νkc(t)+Htc2νkt(t))(x,v)]L(B(0,r))=0, (4.22)
    limsk1limpk1vφ(x,v)[(Hct11νpk(t)+Htt11νsk(t)+Hct2νpk(t)+Htt2νsk(t))(x,v)(Hct11νkc(t)+Htt11νkt(t)+Hct2νkc(t)+Htt2νkt(t))(x,v)]L(B(0,r))=0, (4.23)

    which implies equations (4.18) and (4.19). Now let IkP2 be the set of indices of cars on lane k performing lane-change at least once during the time interval [0,T]. Assume that |IkP2|=pk2. By the lane-changing conditions (2.11) and (2.12), we consider the following discrete measure to track the positions and velocities of these cars:

    νpk2(t)==iIk1P2mcδ(xi(t),vi(t))p1([ˉakc,iak1iΔcc]+,[ˉakt,iak1iΔtc]+,[ˉaki+Δc]+,[ˉakikF+Δc]+,[ˉakikF+Δt]+)(1δ1(k))iIkP2mcδ(xi(t),vi(t))p1([ˉak1c,iakiΔcc]+,[ˉak1t,iakiΔtc]+,[ˉak1i+Δc]+,[ˉak1ik1F+Δc]+,[ˉak1ik1F+Δt]+)(1δ1(k))+iIk+1P2mcδ(xi(t),vi(t))p1([ˉakc,iak+1iΔcc]+,[ˉakt,iak+1iΔtc]+,[ˉaki+Δc]+,[ˉakikF+Δc]+,[ˉakikF+Δt]+)(1δN(k))iIkP2mcδ(xi(t),vi(t))p1([ˉak+1c,iakiΔcc]+,[ˉak+1t,iakiΔtc]+,[ˉak+1i+Δc]+,[ˉak+1ik+1F+Δc]+,[ˉak+1ik+1F+Δt]+)(1δN(k))

    where for j{1,,N} and j=j+1 or j1. The accelerations aji, ˉajc,i and ˉajt,i are respectively given by

    aji=˙vji=(Hcc11νpj+Htc11νsj+Hcc2νpj+Htc2νsj)(xi,vi)iIP,
    ˉajc,i=(Hcc11νpj+Htc11νsj)(xi,vi)+(Hcc2νpj+Htc2νsj)(xi,vi)iIjP,
    ˉajt,i=(Hct11νpj+Htt11νsj)(xi,vi)+(Hct2νpj+Htt2νsj)(xi,vi)iIjP.

    Similarly, for trucks, let IkS2 be the set of indices of trucks on lane k that perform lane-changing at least once during the time interval [0,T]. Denote the number of these trucks by sk2, i.e., |IkS2|=sk2. Again, by the lane-changing conditions (2.11) and (2.12), we consider the following discrete measure to track the positions and velocities of these trucks:

    νsk2(t)==iIk1S2mtδ(xi(t),vi(t))p2([ˉakc,iak1iΔct]+,[ˉakt,iak1iΔtt]+,[ˉaki+Δt]+,[ˉakikF+Δc]+,[ˉakikF+Δt]+)(1δ1(k))iIkS2mtδ(xi(t),vi(t))p2([ˉak1c,iakiΔct]+,[ˉak1t,iakiΔtt]+,[ˉak1i+Δt]+,[ˉak1ik1F+Δc]+,[ˉak1ik1F+Δt]+)(1δ1(k))+iIk+1S2mtδ(xi(t),vi(t))p2([ˉakc,iak+1iΔct]+,[ˉakt,iak+1iΔtt]+,[ˉaki+Δt]+,[ˉakikF+Δc]+,[ˉakikF+Δt]+)(1δN(k))iIkS2mtδ(xi(t),vi(t))p2([ˉak+1c,iakiΔct]+,[ˉak+1t,iakiΔtt]+,[ˉak+1i+Δt]+,[ˉak+1ik+1F+Δc]+,[ˉak+1ik+1F+Δt]+)(1δN(k))

    with j{1,,N}, j=j+1 or j1 and again the accelerations are given by

    aji=˙vji=(Hct11νpj+Htt11νsj+Hct2νpj+Htt2νsj)(xi,vi)iIS,
    ˉajc,i=(Hcc11νpj+Htc11νsj)(xi,vi)+(Hcc2νpj+Htc2νsj)(xi,vi)iIjP
    ˉajt,i=(Hct11νpj+Htt11νsj)(xi,vi)+(Hct2νpj+Htt2νsj)(xi,vi)iIjS.

    Now on each lane k we let the number of cars and trucks performing lane-chance (pk2, sk2) go to infinity. Then we have

    limpk2limsk2νpk2(t)==νk1cp1([AkPAk1PΔcc]+,[AkSAk1PΔtc]+,[AkP+Δc]+,[AkP+Δc]+,[AkS+Δt]+)(1δ1(k))νkcp1([Ak1PAkPΔcc]+,[Ak1SAkPΔtc]+,[Ak1P+Δc]+,[Ak1P+Δc]+,[Ak1S+Δt]+)(1δ1(k))+νk+1cp1([AkPAk+1PΔcc]+,[AkSAk+1PΔtc]+,[AkP+Δc]+,[AkP+Δc]+,[AkS+Δt]+)(1δN(k))νkcp1([Ak+1PAkPΔcc]+,[Ak+1SAkPΔtc]+,[Ak+1P+Δc]+,[Ak+1P+Δc]+,[Ak+1S+Δt]+)(1δN(k))=[Gk1,k1(νk1c,νk1t,νkc,νkt)Gk,k11(νkc,νkt,νk1c,νk1t)](1δ1(k))+[Gk+1,k1(νk+1c,νk+1t,νk1c,νk1t)Gk,k+11(νkc,νkt,νk+1c,νk+1t)](1δN(k))=G1(νkc,νkt,νkc,νkt) (4.24)
    limsk2limpk2νsk2(t)==νk1tp2([AkPAk1SΔct]+,[AkSAk1SΔtt]+,[AkS+Δt]+,[AkP+Δc]+,[AkP+Δt]+)(1δ1(k))νktp2([Ak1PAkSΔct]+,[Ak1SAkSΔtt]+,[Ak1S+Δt]+,[Ak1P+Δc]+,[Ak1S+Δt]+)(1δ1(k))+νk+1tp2([AkPAk+1SΔct]+,[AkSAk+1SΔtt]+,[AkS+Δt]+,[AkP+Δc]+,[AkS+Δt]+)(1δN(k))νktp2([Ak+1PAkSΔct]+,[Ak+1SAkSΔtt]+,[Ak+1S+Δt]+,[Ak+1P+Δc]+,[Ak+1S+Δt]+)(1δN(k))=[Gk1,k2(μk1P,νk1t,νkc,νkt)Gk,k12(νkc,νkt,μk1P,νk1t)](1δ1(k))+[Gk+1,k2(νk+1c,νk+1t,νk1c,νk1t)Gk,k+12(νkc,νkt,νk+1c,νk+1t)](1δN(k))=G2(νkc,νkt,νkc,νkt) (4.25)

    Set νkc=limpk1limpk2(νpk1+νpk2) and νkt=limsk1limsk2(νsk1+νsk2). By combining Eqs (4.16), (4.18), (4.24) and (4.17), (4.19), (4.25) we can observe thath the constructed couple (νkc,νkt), k{1,,N}Mac0(D)M1(D) is a weak solution to the coupled PDEs Section 4 to Eq (4.3).

    Theorem 4.2 (Continuity with respect to the initial conditions). For q=1,2, let (νqc,νqt)(Mac0(D)M1(D))2L be two weak solutions for the coupled equations (4.2), (4.3) over the time interval [0,T] associated to the initial data (νq0,c,νq0,t)(Mac0(D)M1(D))2L. Then for all k{1,,L}, there exists a positive constant C0 such that

    W1,11(νk,1c(t),νk,2c(t))+W1,11(νk,1t(t),νk,2t(t))C0Li=1(W1,11(νi,10,c,νi,20,c)+W1,11(νi,10,t,νi,20,t)),t[0,T]. (4.26)

    Here we have assumed for each q=1,2, νq0,c=(νk,q0,c)Lk=1 and the same for νqc,νqt and νq0,t.

    Proof. Let (νk,qc,νk,qt):[0,T](Mac0(D)M1(D))2 be two solutions to system (4.2), (4.3) over the time interval [0,T] associated to the initial data (νk,q0,c,νk,q0,t)(Mac0(D)M1(D))2 with q=1,2 and k{1,,L}. Let t[0,T] be fixed and denote Δt=T2j for a fixed jN+. Consider the partition of [0,T] into sub-intervals [0,Δt), [Δt,2Δt),,[(2j1)Δt,2jΔt) and let n be the maximum integer such that tnΔt0, then t[nΔt,(n+1)Δt). As mentioned in Section 2.5, (νk,qc,νk,qt)=limj(νk,qj,c,νk,qj,t), where (νk,qj,c,νk,qj,t) is constructed according to the following scheme:

    (νk,qj,c(0),νk,qj,t(0)):=(νk,q0,c,νk,q0,t);(νk,qj,c((n+1)Δt),νk,qj,t((n+1)Δt)):=Tνk,qj,c(nΔt),νk,qj,t(nΔt)Δt#(νk,qj,c(nΔt),νk,qj,t(nΔt))++Δt(G1(νk,qj,c(nΔt),νk,qj,t(nΔt),νk,qj,c(nΔt),νk,qj,t(nΔt)),G2(νk,qj,c(nΔt),νk,qj,t(nΔt),νk,qj,c(nΔt),νk,qj,t(nΔt)));(νk,qj,c(t),νk,qj,t(t)):=Tνk,qj,c(nΔt),νk,qj,t(nΔt)τ#(νk,qj,c(nΔt),νk,qj,t(nΔt))++τ(G1(νk,qj,c(nΔt),νk,qj,t(nΔt),νk,qj,c(nΔt),νk,qj,t(nΔt)),G2(νk,qj,c(nΔt),νk,qj,t(nΔt),νk,qj,c(nΔt),νk,qj,t(nΔt))).

    with k{k1,k+1} and τ=tnΔt. Observe that for t=(n+1)Δt we have τ=Δt.

    The key observation is that in this approximation procedure the equations in (2.9) for the evolution of different vehicles, are decoupled. The consequence is that the flow Tνk,qj,c(nΔt),νk,qj,t(nΔt)τ has two components (Tk,qc,Tk,qt) representing respectively the evolution of cars and trucks.

    For every t[0,T],

    W1,11(νk,1j,c(t),νk,2j,c(t))==W1,11(Tk,1c#(νk,1j,c(nΔt))+τG1(νk,1j,c(nΔt),νk,1j,t(nΔt),νk,1j,c(nΔt),νk,1j,t(nΔt)),Tk,2c#(νk,2j,c(nΔt))+τG1(νk,2j,c(nΔt),νk,2j,t(nΔt),νk,2j,c(nΔt),νk,2j,t(nΔt)))W1,11(Tk,1c#(νk,1j,c(nΔt)),Tk,2c#(νk,2j,c(nΔt)))+τW1,11(G1(νk,1j,c(nΔt),νk,1j,t(nΔt),νk,1j,c(nΔt),νk,1j,t(nΔt)),G1(νk,2j,c(nΔt),νk,2j,t(nΔt),νk,2j,c(nΔt),νk,2j,t(nΔt))) (4.27)

    Lemma 2.3 together with the estimate Eq (6.14) in [18] yield the existence of a radius ρ>0 and three constants L1,L2,L3>0 such that

    W1,11(Tk,1c#(νk,1j,c(nΔt)),Tk,2c#(νk,2j,c(nΔt)))W1,11(Tk,1c#(νk,1j,c(nΔt)),Tk,1c#(νk,2j,c(nΔt)))+W1,11(Tk,2c#(νk,1j,c(nΔt)),Tk,2c#(νk,2j,c(nΔt)))L1W1,11(νk,1j,c(nΔt),νk,2j,c(nΔt))+L3tnΔteL2(ts)[W1,11(νk,1j,c(s)),νk,2j,c(s))+W1,11(νk,1j,t(s)),νk,2j,t(s))]ds (4.28)

    where in the first passage we applied the triangular inequality and then the Lipschitz continuity of the flow map with an application of the Gronwall's Lemma.

    On the other side the source G1 is Lipschitz continuous in all the input with constant LG1, therefore

    W1,11(G1(νk,1j,c(nΔt),νk,1j,t(nΔt),νk,1j,c(nΔt),νk,1j,t(nΔt)),G1(νk,2j,c(nΔt),νk,2j,t(nΔt),νk,2j,c(nΔt),νk,2j,t(nΔt)))LG1(W1,11(νk,1j,c(nΔt),νk,2j,c(nΔt))+W1,11(νk,1j,t(nΔt),νk,2j,t(nΔt))+W1,11(νk,1j,c(nΔt),νk,2j,c(nΔt))+W1,11(νk,1j,t(nΔt),νk,2j,t(nΔt))) (4.29)

    Combining recursively Eqs (4.27)–(4.29) we find the following estimate:

    W1,11(νk,1j,c(t),νk,2j,c(t))C1Li=1(W1,11(νi,10,c,νi,20,c)+W1,11(νi,10,t,νi,20,t)) (4.30)

    with C1 positive constant. In analogous way we can derive

    W1,11(νk,1j,t(t),νk,2j,t(t))C2Li=1(W1,11(νi,10,c,νi,20,c)+W1,11(νi,10,t,νi,20,t)) (4.31)

    for C2>0. By adding Eq (4.30) to Eq (4.31) and taking the limit for j we obtain Eq (4.26) with C0=max{C1,C2}.

    In the future, one may study the dynamics of finitely many vehicles including cars and trucks on a multi-lane in an appropriate numerical scheme. In particular, the parameters for the Bando-Follow-the-leader model and the lane-changing probability functions are needed to be trained. Furthermore, the convergence of the finite-dimensional hybrid system to the Vlasov type PDE with a source term can also be studied numerically. In addition, one may also add the dynamics of finitely many controlled autonomous vehicles to the study and focus on an optimal control problem to minimize, for instance, energy cost, and so on. In that case, we expect to have the convergence of a finite-dimensional hybrid optimal control problem to an infinite-dimensional hybrid optimal control problem.

    The research of X. G. was partially supported by the NSF CPS Synergy project "Smoothing Traffic via Energy-efficient Autonomous Driving" (STEAD) CNS 1837481.

    The research of B. P. is based upon work supported by the U.S. Department of Energy's Office of Energy Efficiency and Renewable Energy (EERE) under the Vehicle Technologies Office award number CID DE-EE0008872. The views expressed herein do not necessarily represent the views of the U.S. Department of Energy or the United States Government.

    The authors declare there is no conflict of interest.



    [1] G Albi, N Bellomo, L Fermo, S. Y Ha, J. Kim, L. Pareschi, D. Poyato, J. Soler, Vehicular traffic, crowds, and swarms: From kinetic theory and multiscale methods to applications and research perspectives, Math Models Methods Appl Sci, 29 (2019): 1901–2005. https://doi.org/10.1142/S0218202519500374 doi: 10.1142/S0218202519500374
    [2] L. Ambrosio, N. Gigli, G. Savaré, Gradient flows: in metric spaces and in the space of probability measures, Berlin: Springer Science & Business Media, 2008.
    [3] M. Bando, K. Hasebe, A. Nakayama, A. Shibata, Y. Sugiyama. Structure stability of congestion in traffic dynamics, Jpn J Ind Appl Math, 11 (1994), 203–223. https://doi.org/10.1007/BF03167222 doi: 10.1007/BF03167222
    [4] N. Bellomo, C. Dogbe, On the modeling of traffic and crowds: A survey of models, speculations, and perspectives, SIAM Rev Soc Ind Appl Math, 53 (2011), 409–463. https://doi.org/10.1137/090746677 doi: 10.1137/090746677
    [5] S. Benzoni-Gavage, R. M. Colombo, An n-populations model for traffic flow, Eur J Appl Math, 14 (2003), 587–612.
    [6] V. I. Bogachev, Measure Theory, Heidelberg: Springer Berlin, 2007.
    [7] R. Borsche, A. Klar, M. Zanella, Kinetic-controlled hydrodynamics for multilane traffic models, Physica A, 587 (2022), 126486. https://doi.org/10.1016/j.physa.2021.126486 doi: 10.1016/j.physa.2021.126486
    [8] M. S. Branicky, V. S. Borkar, S. K. Mitter, A unified framework for hybrid control: model and optimal control theory, IEEE Trans. Automat. Contr., 43 (1998), 31–45. https://doi.org/10.1109/9.654885 doi: 10.1109/9.654885
    [9] P. Cardaliaguet, N. Forcadel, From heterogeneous microscopic traffic flow models to macroscopic models, SIAM J. Math. Anal., 53 (2021), 309–322. https://doi.org/10.1137/20M1314410 doi: 10.1137/20M1314410
    [10] M. T. Chiri, X. Q. Gong, B. Piccoli, Hybrid multi-population traffic flow model: Optimal control for a mean-field limit, [Preprint], (2021) [cited 2022 Feb 20 ]. Available from: https://cvgmt.sns.it/media/doc/paper/5302/Conference_paper_Car_Truck_with_control.pdf
    [11] V. Coscia, M. Delitala, P. Frasca, On the mathematical theory of vehicular traffic flow Ⅱ: Discrete velocity kinetic models, Int J Non Linear Mech, 42 (2007), 411–421. https://doi.org/10.1016/j.ijnonlinmec.2006.02.008 doi: 10.1016/j.ijnonlinmec.2006.02.008
    [12] I. D. Couzin, J. Krause, N. R. Franks, S. A. Levin, Effective leadership and decision-making in animal groups on the move, Nature, 433 (2005), 513–516. https://doi.org/10.1038/nature03236 doi: 10.1038/nature03236
    [13] E. Cristiani, B. Piccoli, A. Tosin, Multiscale modeling of granular flows with application to crowd dynamics, Multiscale Model Simul, 9 (2011), 155–182. https://doi.org/10.1137/100797515 doi: 10.1137/100797515
    [14] F. Cucker, S. Smale, Emergent behavior in flocks, IEEE Trans. Automat. Contr., 52 (2007), 852–862. https://doi.org/10.1109/TAC.2007.895842 doi: 10.1109/TAC.2007.895842
    [15] M. Delitala, A. Tosin, Mathematical modeling of vehicular traffic: a discrete kinetic theory approach, Math. Models Methods Appl. Sci., 17 (2007), 901–932. https://doi.org/10.1142/S0218202507002157 doi: 10.1142/S0218202507002157
    [16] M. L. D. Monache, T. Liard, Anaïs Rat, R. Stern, R. Bhadani, B. Seibold, et al., Feedback Control Algorithms for the Dissipation of Traffic Waves with Autonomous Vehicles, Cham: Springer International Publishing, 2019,275–299.
    [17] A. Festa, S. Göttlich, A mean field game approach for multi-lane traffic management, IFAC-PapersOnLine, 51 (2018), 793–798.
    [18] M. Fornasier, B. Piccoli, F. Rossi, Mean-field sparse optimal control, Philos. Trans. Royal Soc., 372 (2014), 20130400.
    [19] M. Garavello, B. Piccoli, Hybrid necessary principle, SIAM J Control Optim, 43 (2005), 1867–1887. https://doi.org/10.1137/S0363012903416219 doi: 10.1137/S0363012903416219
    [20] D. C. Gazis, R. Herman, R. W. Rothery, Nonlinear follow-the-leader models of traffic flow, Oper. Res., 9 (1961), 545–567. https://doi.org/10.1287/opre.9.4.545 doi: 10.1287/opre.9.4.545
    [21] R. Goebel, R. G. Sanfelice, A. R. Teel, Hybrid dynamical systems, IEEE Control Syst. Mag., 29 (2009), 28–93. https://doi.org/10.1109/MCS.2008.931718 doi: 10.1109/MCS.2008.931718
    [22] X. Q. Gong, A. Keimer, On the well-posedness of the "bando-follow the leader" car following model and a "time-delayed version", [Preprint], (2022) [cited 2022 Feb 24 ]. Available from: 10.13140/RG.2.2.22507.62246
    [23] X. Q. Gong, B. Piccoli, G. Visconti, Mean-field limit of a hybrid system for multi-lane multi-class traffic, arXiv: 2007.14655, [Preprint], (2020) [cited 2022 Feb 24 ]. Available from: https://doi.org/10.48550/arXiv.2007.14655
    [24] Jack K. Hale, Ordinary differential equations, Roberte E. New York: Krieger Publishing company, 1980.
    [25] M Herty, R Illner, A Klar, V Panferov, Qualitative properties of solutions to systems of fokker-planck equations for multilane traffic flow, Transport Theor Stat Phys, 35 (2006), 31–54. https://doi.org/10.1080/00411450600878573 doi: 10.1080/00411450600878573
    [26] M. Herty, S. Moutari, G. Visconti, Macroscopic modeling of multilane motorways using a two-dimensional second-order model of traffic flow, SIAM J. Appl. Math., 78 (2018), 2252–2278. https://doi.org/10.1137/17M1151821 doi: 10.1137/17M1151821
    [27] M. Herty, L. Pareschi, Fokker-Planck asymptotics for traffic flow models, Kinet. Relat. Models, 3 (2010), 165–179.
    [28] M. Herty, G. Puppo, G. Visconti, Model of vehicle interactions with autonomous cars and its properties, arXiv: 2107.14081, [Preprint], (2021) [cited 2022 Feb 20 ]. Available from: https://doi.org/10.48550/arXiv.2107.14081
    [29] M. Herty, G. Visconti, Analysis of risk levels for traffic on a multi-lane highway, IFAC-PapersOnLine, 51 (2018), 43–48.
    [30] H. Holden, N. H. Risebro, Models for dense multilane vehicular traffic, SIAM J. Math. Anal., 51 (2019), 3694–3713. https://doi.org/10.1137/19M124318X doi: 10.1137/19M124318X
    [31] R. Illner, A. Klar, T. Materne, Vlasov-Fokker-Planck models for multilane traffic flow, Commun. Math. Sci., 1 (2003), 1–12.
    [32] N. Kardous, A. Hayat, S. McQuade, X. Q. Gong, S. Truong, P. Arnold, et al., A rigorous multi-population multi-lane hybrid traffic model and its mean-field limit for dissipation of waves via autonomous vehicles, arXiv: 2205.06913, [Preprint], (2021) [cited 2022 Feb 20 ]. Available from: https://doi.org/10.48550/arXiv.2205.06913
    [33] A. Kesting, M. Treiber, D. Helbing, General lane-changing model mobil for car-following models, Trans Res Rec, 1999 (2007), 86–94. https://doi.org/10.3141/1999-10 doi: 10.3141/1999-10
    [34] A. Klar, R. Wegener, Enskog-like kinetic models for vehicular traffic, J. Stat. Phys., 87 (1997), 91–114. https://doi.org/10.1007/BF02181481 doi: 10.1007/BF02181481
    [35] S. L. Paveri-Fontana, On Boltzmann-like treatments for traffic flow: a critical review of the basic model and an alternative proposal for dilute traffic analysis, Trans. Res., 9 (1975), 225–235. https://doi.org/10.1016/0041-1647(75)90063-5 doi: 10.1016/0041-1647(75)90063-5
    [36] B. Piccoli, A. Tosin, M. Zanella, Model-based assessment of the impact of driver-assist vehicles using kinetic theory, Z Angew Math Phys, 71 (2020), 1–25. https://doi.org/10.1007/s00033-019-1224-x doi: 10.1007/s00033-019-1224-x
    [37] B. Piccoli, Hybrid systems and optimal control, Proceedings of the 37th IEEE Conference on Decision and Control, 1 (1998), 13–18.
    [38] B. Piccoli, F. Rossi, Generalized wasserstein distance and its application to transport equations with source, Arch Ration Mech Anal, 211 (2014), 335–358. https://doi.org/10.1007/s00205-013-0669-x doi: 10.1007/s00205-013-0669-x
    [39] B. Piccoli, A. Tosin, Vehicular traffic: A review of continuum mathematical models, Encycl. Complex. Syst. Sci., 22 (2009), 9727–9749.
    [40] I. Prigogine, A Boltzmann-like approach to the statistical theory of traffic flow, In: R. Herman, editor, Theory of traffic flow, Amsterdam: Elsevier, 1961,158–164.
    [41] I. Prigogine, R. Herman, Kinetic theory of vehicular traffic, New York: American Elsevier Publishing, 1971.
    [42] G. Puppo, M. Semplice, A. Tosin, G. Visconti, Analysis of a multi-population kinetic model for traffic flow, Commun Math Sci, 15 (2017), 379–412. https://doi.org/10.4310/CMS.2017.v15.n2.a5 doi: 10.4310/CMS.2017.v15.n2.a5
    [43] A Reuschel, Vehicle movements in a platoon with uniform acceleration or deceleration of the lead vehicle, Zeitschrift des Oesterreichischen Ingenieur-und Architekten-Vereines, 95 (1950), 50–62.
    [44] J. Song, S. Karni, A second order traffic flow model with lane changing, J. Sci. Comput., 81 (2019), 1429–1445. https://doi.org/10.1007/s10915-019-01023-z doi: 10.1007/s10915-019-01023-z
    [45] Raphael E. Stern, S. Cui, Maria Laura Delle Monache, R. Bhadani, M. Bunting, M. Churchill, et al., Dissipation of stop-and-go waves via control of autonomous vehicles: Field experiments, Transp Res Part C Emerg Technol, 89 (2018), 205–221. https://doi.org/10.1016/j.trc.2018.02.005 doi: 10.1016/j.trc.2018.02.005
    [46] A. B. Sukhinova, M. A. Trapeznikova, B. N. Chetverushkin, N. G. Churbanova, Two-dimensional macroscopic model of traffic flows, Math. Models Comput. Simul., 1 (2009), 669–676. https://doi.org/10.1134/S2070048209060027 doi: 10.1134/S2070048209060027
    [47] C. Tomlin, G. J. Pappas, S. Sastry, Conflict resolution for air traffic management: a study in multiagent hybrid systems, IEEE Trans. Automat. Contr., 43 (1998), 509–521. https://doi.org/10.1109/9.664154 doi: 10.1109/9.664154
    [48] T. Trimborn, L. Pareschi, M. Frank, Portfolio optimization and model predictive control: A kinetic approach, Discrete Cont. Dyn.-B, 24 (2019), 6209–6238. https://doi.org/10.3934/dcdsb.2019136 doi: 10.3934/dcdsb.2019136
    [49] P Zhang, R. X Liu, S. C. Wong, S. Q Dai, Hyperbolicity and kinematic waves of a class of multi-population partial differential equations, Eur J Appl Math, 17 (2006), 171–200. https://doi.org/10.1017/S095679250500642X doi: 10.1017/S095679250500642X
  • This article has been cited by:

    1. Matteo Piu, Michael Herty, Gabriella Puppo, Derivation and Stability Analysis of a Macroscopic Multilane Model for Traffic Flow, 2023, 83, 0036-1399, 2052, 10.1137/22M1543288
    2. Xiaoqian Gong, Michael Herty, Benedetto Piccoli, Giuseppe Visconti, Crowd Dynamics: Modeling and Control of Multiagent Systems, 2023, 6, 2573-5144, 261, 10.1146/annurev-control-060822-123629
  • Reader Comments
  • © 2023 the Author(s), licensee AIMS Press. This is an open access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0)
通讯作者: 陈斌, bchen63@163.com
  • 1. 

    沈阳化工大学材料科学与工程学院 沈阳 110142

  1. 本站搜索
  2. 百度学术搜索
  3. 万方数据库搜索
  4. CNKI搜索

Metrics

Article views(1554) PDF downloads(43) Cited by(2)

Other Articles By Authors

/

DownLoad:  Full-Size Img  PowerPoint
Return
Return

Catalog