The review scope is defined along three dimensions: transportation domain, user role, and system decision level. The application domains include transportation electrification, urban logistics, shared mobility, multimodal transport, and Mobility-as-a-Service (MaaS).

The review distinguishes three user roles, passive demand, operational agents, and system co-shapers, which correspond broadly to demand representation, operational decision-making, and long-term system planning, respectively.

A structured literature search was conducted using Scopus and Web of Science, supplemented by Google Scholar and backward and forward citation tracking. The primary search period covered studies published between 2010 and 2026, while earlier foundational contributions on travel demand modeling, user equilibrium, and discrete choice were retained to establish the theoretical background. Search terms combined user-related concepts such as “user behavior,” “user preference,” “willingness to pay,” “service acceptance,” and “user feedback” with system-level concepts related to transportation electrification, shared mobility, MaaS, multimodal transport, urban logistics, pricing, facility location, and transportation planning.

Studies were retained when they explicitly analyzed or modeled user-related factors and linked them to transportation system decisions or outcomes. Studies focusing solely on individual psychology, driving cognition, traffic safety behavior, vehicle control, battery technology, or supply-side optimization were excluded unless they established a direct connection with transportation system design, operation, or planning.

The selected studies were classified according to transportation domain, user role, and system decision level, with particular attention to how user-related information entered the decision process, as exogenous demand inputs, endogenous behavioral responses, or feedback signals for longer-term planning. This classification provided the basis for developing and refining the role-based taxonomy presented in Section 2.2.

The resulting role-based taxonomy distinguishes three user roles according to how user-related information enters transportation decision models.

In passive-demand studies, users are represented primarily through exogenous parameters, observed states, or baseline constraints that support forecasting, assignment, resource allocation, and capacity planning but provide limited feedback to system decisions.

In operational-agent studies, users enter the decision process through endogenous choices, response functions, utility terms, acceptance constraints, participation decisions, or strategic interactions. Users may change their travel, charging, delivery, or service choices in response to system attributes, and these responses directly affect short-term outcomes such as congestion, occupancy, waiting time, routing efficiency, energy demand, and operator revenue.

In system co-shaping studies, user behavior, feedback, and associated operational outcomes are further translated into planning signals. Repeated congestion, low utilization, rejected offers, service non-use, location preferences, adoption patterns, and complaints may influence infrastructure location, capacity expansion, pricing structures, service coverage, and technology deployment. Explicit user participation through co-design, participatory planning, or living labs represents one form of co-shaping, but the concept also includes indirect influence through observed behavior and market feedback.

The three roles do not describe mutually exclusive user groups or a necessary progression followed by every user. Instead, they represent different ways of using user-related information in transportation decision-making. The same user may simultaneously generate demand, respond to operational conditions, and provide signals that influence future system design.

Figure 1 presents the decision-integration framework used to organize the review. It connects user-related information, role representation, decision use, observed outcomes, and feedback to future decisions. The three horizontal layers correspond to passive demand, operational agents, and system co-shapers, while the vertical direction indicates increasing user influence and integration into transportation decision-making.

Pan/Zoom

Across these layers, user-related information progresses from demand inputs to endogenous operational responses and, ultimately, to feedback signals for infrastructure planning and service redesign. Decisions implemented at each level subsequently alter the conditions experienced by users, creating feedback between users and transportation systems. This framework structures the remainder of the review.

In passenger transportation, users are commonly represented through trip generation, origin–destination matrices, mode-specific demand, and network flows, which underpin conventional travel demand modeling, planning, and traffic assignment [3]. Activity-based models provide a more disaggregate representation by linking travel to daily activities, schedules, and locations, although their outputs are still commonly translated into structured demand inputs for system-level planning and simulation [4].

In freight and urban logistics, customers are represented through order volumes, delivery locations, service time windows, and spatial demand distributions to support routing, facility location, and last-mile network design [5]. Users therefore influence the system primarily through the delivery requirements they generate rather than through responses to prices, locations, or service alternatives.

In transportation electrification, users are commonly represented through charging profiles, arrival and parking patterns, state of charge, required energy, and spatial charging demand. Such information supports charging-infrastructure planning and load management [28, 6], while users remain primarily sources of charging demand to be forecasted or allocated.

Shared mobility and demand-responsive services use comparable representations, including trip requests, pickup–drop-off patterns, station-level demand, and lost demand, to support fleet planning, service coverage, and rebalancing [7].

Across these domains, increasingly detailed observations may improve demand estimation without necessarily changing the passive role assigned to users in subsequent decision models.

Passive-demand models remain essential for large-scale planning but have three main limitations. First, they often treat demand as fixed or only weakly responsive to system conditions, although users may change departure time, route, mode, charging location, delivery option, or service participation in response to price, waiting time, reliability, accessibility, and service availability. Second, aggregated demand can obscure heterogeneity in flexibility, willingness to pay, technology access, service expectations, and individual constraints. Third, passive-demand representations provide limited feedback to system design: they may indicate where and when demand occurs but not whether users will accept a proposed price, schedule, facility, or service configuration. These limitations become increasingly important in platform-mediated and demand-responsive systems and motivate representations in which users enter the decision process as active operational agents.

Behavioral responses make realized demand endogenous to service conditions. EV users, for example, choose when, where, and how quickly to charge, with substantial heterogeneity in charging frequency, regularity, and flexibility [29, 10]. Charging-location decisions are shaped by accessibility, detour distance, waiting time, and surrounding activities, while prices, parking fees, and incentives can alter charging duration, station choice, and temporal demand [30, 9].

The same mechanism appears in other mobility and logistics services. MaaS users respond to bundle composition, price, travel habits, and service attributes through subscription, booking, and mode-combination decisions; both stated-choice studies and real-world trials show that service uptake depends on service design and user characteristics [11, 13]. Last-mile customers similarly respond to delivery cost, time-window availability, convenience, and pickup-point accessibility when choosing among home delivery, service points, and parcel lockers [14, 15]. These choices directly influence realized demand and the operational conditions faced by service providers.

Operational agency becomes stronger when users not only respond to service conditions but also provide flexibility that can be exploited in system operation. Smart charging, demand response, and vehicle-to-grid services provide a clear example: their technical benefits depend on users’ willingness to shift charging or participate in coordinated charging. Optimization studies often treat such participation as given [25, 31], whereas behavioral evidence shows that participation depends on compensation, perceived control, trust, charging reliability, and mobility requirements [8].

Similar forms of flexibility appear in other services. Incentives can encourage electric-carsharing users to accept alternative pickup or drop-off locations and thereby support vehicle relocation [32]. In bike-sharing systems, users willing to walk to nearby stations when their preferred station is empty can reduce lost demand and support rebalancing decisions [33]. In urban logistics, customer flexibility between home delivery and shared locations can improve routing when appropriate compensation is provided [34], while passengers participating in public-transport-based crowdshipping become direct contributors to logistics operations, with participation shaped by compensation, additional time, and task characteristics [35]. Across these cases, user flexibility becomes an operational resource, but its value depends on users’ willingness to participate under the conditions offered.

A key challenge is to move from analyzing user responses to incorporating them directly into operational decisions. Behavioral studies can identify preferences, willingness to pay, acceptance, and participation, while optimization models often simplify these responses or treat them as exogenous assumptions. This separation is evident in flexible charging, where participation and behavioral constraints may determine whether technically optimal schedules can actually be implemented. More generally, predicting whether users will accept a price, service configuration, or operational request does not by itself ensure that such responses influence the decisions made by the operator.

Some studies increasingly bridge this gap by jointly representing user behavior and operational decisions. In EV charging, bid-informed frameworks can embed user-specified time windows, price bids, and energy requirements directly into operator pricing and scheduling decisions [36]. In urban logistics, pricing and self-pickup incentives can be linked with customer responses and delivery operations [37], while more integrated frameworks jointly model delivery alternatives, pricing, customer choice, and routing [38]. Such approaches shift user-related information from a descriptive or predictive input toward an endogenous component of operational decision-making. However, this integration remains uneven across transportation domains, and short-term user responses are rarely translated systematically into subsequent infrastructure or service-planning decisions. This limitation motivates the system co-shaping perspective examined in Section 5.

Operational interactions can generate planning signals when their effects persist or recur over time. Repeated charging congestion may indicate insufficient capacity; persistent rejection of service offers may reveal a mismatch between pricing, scheduling, and user constraints; high utilization may support facility expansion; and low adoption may indicate accessibility, pricing, or service-design problems. In this way, behavioral data, expressed preferences, utilization, acceptance, and broader market responses can inform facility location, capacity expansion, service redesign, investment prioritization, and technology deployment. The key shift is therefore from treating operational outcomes as end-point performance measures to using them as inputs for subsequent planning decisions.

User-related information becomes co-shaping when it is translated into infrastructure or service-planning decisions. Charging infrastructure provides a clear example. User location preferences, detour tolerance, travel patterns, and charging demand can inform station siting and capacity decisions. Ouyang and Xu [1], for example, incorporate detour tolerance and travel cost into charging-station siting, while Liagkas et al. [2] jointly optimize charger placement and dynamic spatio-temporal pricing while allowing price-sensitive users to shift charging location and time or opt out. Utilization and adoption provide additional feedback for infrastructure planning. EV adoption and charging infrastructure are mutually dependent, as insufficient infrastructure may discourage adoption while a limited user base may reduce investment incentives [39]. Local EV penetration and charging patterns are also important predictors of public charging-station utilization [18], indicating how observed user behavior can inform future expansion and capacity decisions.

Similar mechanisms appear in shared mobility and mobility-hub planning. Repeated subscription behavior, willingness to pay, service uptake, and accessibility needs can inform service and infrastructure design, particularly when user responses differ across population groups [21, 22]. Mobility-hub studies further show how accessibility, transfer quality, walking distance, information provision, and other user requirements can shape hub design and location decisions [19, 20]. Accessibility-based planning can complement such evidence by translating commuting patterns, activity destinations, and existing public transport into mobility-hub location and service decisions [40].

User responses can also influence the design of urban logistics networks. Choice-based facility-location studies show that willingness to use self-collection services can affect pickup-point and parcel-locker deployment [23, 24]. Customer flexibility, self-pickup incentives, and delivery-location choices can further affect routing performance and facility utilization, providing planning signals for delivery-network design and shared delivery infrastructure [34, 37]. Across these applications, the central shift is from using user behavior only to describe demand to using behavioral and operational responses as inputs to infrastructure and service design.

Observed behavior can provide valuable planning signals, but it should not automatically be interpreted as unconstrained user preference. Low use, for example, may reflect low demand, but it may also result from poor accessibility, limited information, exclusion, or unreliable service. Similarly, observed users represent only those able and willing to participate, while non-users and latent or constrained demand remain less visible. Translating operational feedback into planning decisions therefore requires distinguishing among preference, acceptance, access, and constraint.

A stronger form of co-shaping complements observed behavior with explicitly expressed user information. In charging services, for example, acceptable charging windows, minimum energy requirements, price limits, charging-speed preferences, and willingness to provide flexibility can reveal requirements that are difficult to infer from utilization alone [36]. Acceptance and rejection of service offers can likewise indicate mismatches between user needs and existing pricing, capacity, or service conditions. Comparable signals can arise from repeated service use or non-use, willingness to pay, accessibility needs, and acceptance of alternative service configurations across mobility and logistics applications.

The distinction between implicit and active co-shaping therefore lies in whether user-related information is systematically interpreted and fed back into subsequent planning decisions. Evidence of such continuous feedback remains limited: most studies rely on stated preferences, observed usage data, or simulation-based planning, while relatively few demonstrate how repeated user responses lead to updated infrastructure or service decisions. This gap points to the need for stronger feedback loops between users, operations, and long-term planning.

Regulatory considerations concern data protection, automated decision-making, and the use of personal information in transportation services. Mobility traces, app interactions, payment records, charging histories, and other user-related data may face restrictions on collection, reuse, storage, and sharing [47, 48]. Automated decisions involving personalized pricing, routing, charging allocation, or delivery-slot assignment may also raise concerns regarding transparency, explainability, consent where applicable, and mechanisms for user contestation. Regulation therefore affects both what user information can be used and how resulting decisions are implemented and communicated.

Market and platform conditions shape cross-provider coordination. User-informed services often depend on interoperability and data sharing, but these may be constrained by fragmented standards, limited willingness to open data access, and competing commercial interests [49]. Dynamic or personalized pricing may also face scrutiny when mechanisms are perceived as opaque, discriminatory, or anti-competitive. Platform governance and market structure therefore influence deployability alongside technical performance.

Societal considerations include acceptance, fairness, accessibility, trust, and public value. Efficiency-oriented solutions may fail when users perceive decisions as unfair, unreliable, difficult to understand, or insufficiently under their control. Pricing and allocation mechanisms may impose greater burdens on users with limited flexibility, while digital-only services may exclude users with limited digital access. Highly personalized services may also be less responsive to users with sparse behavioral data. User-informed transportation systems should therefore consider equity, accessibility, digital inclusion, perceived control, and trust alongside efficiency [50, 51].

Across these dimensions, important implementation gaps remain. Acceptance and trust are rarely represented directly in optimization models; transparency is often acknowledged without being translated into interpretable or auditable decision rules; privacy–performance trade-offs remain insufficiently quantified; and market and platform governance are seldom incorporated explicitly into transportation models. Addressing these gaps requires privacy-preserving and interpretable methods, fairness and acceptance criteria, and market-aware models that account for pricing, data sharing, interoperability, and platform coordination.

Translating user-related information into deployable decision support requires behavioral variables that are observable, operationally relevant, and compatible with existing systems. In practice, this transition may begin with user-informed decision support, where user information improves planning and operational decisions before fully adaptive user-driven systems become feasible.

7.2.1 Available user data and behavioral information gaps

7.2.2 Passenger and commercial transportation requirements

7.2.3 Operational and user-centric performance metrics

7.2.4 Practical considerations and cautions for deployment

A first direction is to move from behavioral prediction to decision integration. Many studies can quantify user preferences, such as charging flexibility, willingness to pay, delivery preference, MaaS bundle preference, and service acceptance [8, 13, 14, 15, 22]. However, these variables are still often used mainly to explain or predict user choices. Future research should further examine how these variables can be embedded into decision models as constraints, objectives, choice probabilities, acceptance thresholds, or feedback signals.

This shift is important because user-related variables can become decision inputs rather than only behavioral outputs. For example, acceptable charging time slots, minimum energy requirements, delivery flexibility, pickup-point acceptance, and MaaS subscription patterns can be embedded into charging scheduling, dynamic pricing, delivery routing, bundle design, and capacity expansion models. The key question is how to design models that are both behaviorally realistic and operationally usable.

A second direction is to move from implicit behavioral feedback to active user participation. Current studies often infer preferences from observed service use, but such behavior may be ambiguous: non-use, for example, may reflect high price, poor accessibility, low reliability, limited information, lack of trust, or unavailable service rather than low demand.

Future systems should provide mechanisms through which users can explicitly express preferences, constraints, and willingness to participate. This is particularly important for services that depend on user cooperation, such as smart charging, vehicle-to-grid, self-pickup delivery, crowdshipping, and user-based relocation in shared mobility [8, 32, 35, 37]. In such settings, user participation should be treated as a design input rather than only an observed outcome.

A third direction is to develop closed-loop frameworks linking user responses, operational outcomes, and long-term planning. Existing planning models already consider utilization, demand, accessibility, and service coverage [28, 40, 1, 23], but user feedback is often represented indirectly or statically.

Future research should examine how repeated operational outcomes can become planning signals. Charging congestion may indicate a need for capacity expansion; low acceptance of smart-charging schedules may reveal inadequate compensation or reliability; and low parcel-locker adoption or MaaS renewal may indicate accessibility, trust, or service-design problems. User–system interaction should therefore be modeled as a continuous feedback process through which operational experience informs subsequent planning and service adaptation.

A fourth direction is to better represent user heterogeneity and strategic behavior. Users differ in value of time, price sensitivity, flexibility, accessibility constraints, charging access, digital access, trust, and service expectations, and such differences can materially affect system outcomes [29, 11, 10, 21, 24].

Future models should avoid relying only on representative users or average preferences when heterogeneity affects system outcomes. User segmentation, mixed logit models, latent class models, agent-based models, and data-driven clustering can support more differentiated service design. Strategic behavior also deserves greater attention, particularly in systems with dynamic pricing, priority rules, incentives, or compensation, where users may adjust their stated flexibility, urgency, or willingness to participate. Incentive-compatible and fairness-aware mechanisms are therefore needed to account for such responses.

A fifth direction is to integrate technical optimization with user acceptance. Many optimization studies focus on cost, travel time, routing efficiency, grid load, or infrastructure utilization, while participation and acceptance are simplified or assumed. Yet the reviewed evidence shows that technically attractive solutions may fail if users perceive them as inconvenient, unreliable, unfair, or insufficiently beneficial [8, 54, 55, 56, 20].

Future optimization models should therefore treat acceptance as part of system feasibility and performance rather than as an exogenous assumption. Acceptance-related constraints, objectives, response functions, or participation probabilities can be used to connect technical decisions with user requirements concerning convenience, reliability, affordability, trust, and perceived control. Behavioral feasibility should thus be evaluated jointly with technical optimality.

A sixth direction is to develop user-driven transportation systems that are adaptive, responsible, and privacy-preserving. Such systems increasingly rely on detailed behavioral and operational data, including mobility traces, transactions, service interactions, and user feedback. While these data can support more responsive pricing, scheduling, planning, and service design, they also raise concerns about privacy, data ownership, consent, and trust [47, 48]. Future research should therefore integrate privacy protection and data governance into system design rather than treating them as external implementation requirements. Privacy-preserving analytics should be accompanied by transparent rules concerning what data are collected, how they are used, and how users can contest or modify data-driven decisions.

Responsible design must also account for equity and public value. Data-driven systems may disproportionately respond to high-demand, high-income, or digitally active users while overlooking users with limited digital access, lower willingness to pay, or constrained mobility options. User-driven design should therefore not become purely market-driven design; accessibility, distributional effects, and fairness should be evaluated alongside efficiency and responsiveness [50, 51].

These concerns become increasingly important as AI-mediated platforms and recommendation systems shape information access and user choices. Users may base travel and service decisions on system-wide information about travel time, prices, charging availability, emissions, reliability, and congestion, potentially improving individual decision quality and system responsiveness [57]. However, AI-mediated recommendations may also reshape demand, concentrate or shift congestion, and steer users toward choices that are individually rational but collectively undesirable [58]. Unequal access to information may introduce additional distributional concerns. Future research should therefore examine how platform-mediated information affects demand formation and system externalities, and how users, platforms, and public authorities should share responsibility for these outcomes.

Table 5 summarizes the six research directions discussed in this section by linking each future direction with a core research question and a key caution.