3.2 Team Trip Planning: Problem and Model Description
3.2.1 Problem Formulation
Consider a group of N travellers, T R = {tr1, tr2, ., tri, ., trN}, in which a traveler tri
is contemplating a trip from an initial location stri to a final destination ftri. Traveler
tri can make the trip by following path ptri ∈ Ptri. Ptri denotes the feasible paths
between stri to ftri that are available to tr
i. There is a certain cost value ctri associated
with each path ptri. ctri is a function of several factors including transportation modals
used to complete the trip (e.g, public vs private), environmental conditions along the path, as well as traffic conditions. The notion of cost in this work is multi-aspect, in the sense that it explicitly quantifies monetary costs, temporal costs, safety costs, and comfort costs, to the extent that a multi-criteria cost formulation is employed to facilitate the trip route optimization process. The concept of doctrine is introduced to capture travellers’ preferences, demands, and constraints. For each path ptri, we use
γtri to denote traveler tr
i’s doctrine for this path. Γtri denotes the set of all doctrines
associated with tri’s feasible paths Ptri. The traveler tri’s desirable route can then
be stated as
ptri
Opt = M in
∀ptri∈Ptri
where Υ represents a selfish selection process by which an optimum path is chosen. ptri
k represents a possible path, and P
tri represents the set of all paths. A selfish
selection process is a process in which the drivers attempt to maximize their benefits regardless of the impact of their chosen actions on the system- negatively or positively. For N travellers, their interaction process and their travelling decisions must be formulated such that optimizing traveller tri’s plan does not negatively impact other
travellers’, tr−i, chosen routes. To deal with these kinds of interactions, we define a
team trip planning game. The team trip planning game, Σ, is a 4-tuple (G, Γtri, stri,
ftri) such that:
• G is a directed acyclic graph that includes all possible routes, Ptri. For each
traveller tri, stri and ftri are defined.
• Γtri is a traveller selfish assessment process which assigns a non-negative value
to each road segment, lpk
j , denoting the cost of this segment.
A solution set of paths Ptri is defined for the game Σ such that each path, ptri
k
connecting stri and ftri, is composed of connected links lp tri k j | l ptrik j ∈ Lp tri k . The cost of these paths is ctri pk = Σ |Lptrik | j=1 l ptrik
j ∗ξr, where ξris a weight value that reflects the degree of
preference that each traveller has for a certain path. For the mobility planning game Σ, traveller tri has an ordered set of strategies: Ptri : Ptri =ptr1i, p
tri 2 , , , , ptrni in which ptri k p tri −k : p tri k = p tri
Opt for tri . Travellers own road segments, l
pk
j s, according
to a mapping function o : Lpk → T R such that o(lpk
j ) = tri (i.e., road segment lpjk is
chosen by traveler trias a possible path). For any path, pk, o(L
pk
S ) is a set of connected
segments satisfying certain travelling criteria and owned by a group of travellers. Suppose that for each traveller tri there is a travelling cost ctrpki and mobility plan-
ning reward Dtri
Rpk. Both cost and reward values are non-negative,
n ctri pk, D tri Rpk ∈ R+ o ,
and they belong to the same currency domain. For example, if the reward is monetary, the cost should be also monetary. For Dtri
Rpk > c tri
pk, traveller tri is able to generate
profit. Furthermore, assume that a group of travellers, i.e., coalition S ⊂ N , through certain agreements, can generate a profit using only paths PS owned by the coalition through certain agreements; for example, the set of paths P is owned by a coalition S if o(P ) ⊂ S. A situation in which there exists a successful team trip plan is called a team trip planning game (N, vσ). vσ is a characteristic function for this game, which
is computed as follows: vσ(S) = ζ(cS P, DRS) if S owns P S ∈ Σ, and ζ(cS P, DRS) > 0 0 otherwise (3.2)
Where ζ is the mapping function which associates every coalition S with a non- negative value according to cS
P and DRS.
The cost of each path, ctri
pk, for traveller tri, is not independent from its reward
value, Dtri
Rpk. Thus, it is necessary to define a positive function, for the coalition S,
that defines the relationship between cSP and DPRS. Furthermore, assume that the
ctri P 6= c tr(−i) P and D tri Rpk 6= D tr(−i)
Rpk . This game should be considered a game with Non-
Transferable Utility (N-TU). In this cooperative game, the solution, i.e., the core, is described as follows: core = ( x ∈ RN|X i∈N xi(N ) = v(N ) and X i∈S xi(S) ≥ v(S) ∀S ∈ 2N \ ∅ ) (3.3) whereP
i∈Nxi(N ) = v(N ) guarantees the efficiency of the outcome, and
P
i∈Sxi(S) ≥
v(S) ∀S ∈ 2N \ ∅ guarantees the stability of the game.
Furthermore, each traveller, tri, has a set of strategies based on which the decision
of the best path, ptri
k , can be obtained. For each traveller, tri, there is a finite non-
u, such that u: Ptri → R. Paths in Ptri are associated with preference relation %
such that u(ptri
i ) ≥ u(p tri j ) if p tri i % p tri
j . Cost functions and rewards can correspond to
utility values, and the ordering of Ptri is conducted through the knowledge of %(tri).
Moreover, Ptri can be expanded to include a non-route related actions. The chosen
departure and the expected arrival times for a trip can be considered as strategic actions, which might change the outcome of the game. For traveller, tri, Ptri can
be ordered according to %(tri) such that ptri
k is better than p tri k+1 iff u(p tri k ) ≥ u(p tri k+1).
This game is described as (N, P, u) game.