In classical logistics and routing problems, resource replenishment is treated as a binary operation: a vehicle visits a depot or a fueling station and fills its capacity to 100%. While this assumption holds for internal combustion vehicles—where refueling takes minutes regardless of tank size—it creates a massive computational and operational bottleneck when applied to electric vehicles (EVs), unmanned aerial vehicles (UAVs), and mobile robotic networks.
Battery replenishment is highly time-consuming. Forcing an electric vehicle to undergo a full recharge at every stop introduces severe operational downtime. Recent research across logistics, multi-agent systems, and fleet management has converged on a more flexible paradigm: partial recharging strategies. By treating a vehicle's state-of-charge (SoC) as a continuous decision variable rather than a binary reset, routing algorithms can optimize both where a vehicle charges and how much energy it takes on.
The Core Intuition: Demand-Aware Energy Matching
Why does partial charging work so well? The fundamental insight is that a vehicle only needs enough energy to safely complete its next operational segment or return to its primary depot. Acquiring any energy beyond that threshold yields diminishing returns or outright operational penalties due to unnecessary waiting times.
As Qi et al. noted in early multi-depot routing frameworks, forcing full replenishment ignores the significant time-saving benefits of taking on only the specific volume needed for the immediate next segment of a route (A generalized multi-depot vehicle routing problem with replenishment based on LocalSolver). In the context of EVs, Zhou et al. demonstrated that calculating the minimum energy required to reach the next destination or return to the depot optimizes operational time and reduces charging fees (The electric vehicle routing problem with partial recharge and vehicle recycling).
Traditional Policy:
[Depot] ---> (Deliveries) ---> [Charger: Wait to 100%] ---> (Deliveries) ---> [Depot]
Partial Recharging Policy:
[Depot] ---> (Deliveries) ---> [Charger: Charge +20% Only] ---> (Deliveries) ---> [Depot]
By decoupling the charging decision from a fixed 100% target, the vehicle minimizes its idle time at the plug. This flexibility is particularly crucial when dealing with strict customer delivery windows or stochastic urban environments. As shown by Nathwani et al., partial recharging strategies allow vehicles to minimize downtime by only acquiring the energy needed for the next leg of a journey, bypassing the rigid constraints of binary replenishment models (Deep Reinforcement Learning for the Electric Vehicle Routing Problem With Time Windows). This capability adds a layer of optimization complexity, but it yields substantial savings in overall trip duration (Hybrid Memetic Search for Electric Vehicle Routing With Time Windows, Simultaneous Pickup-Delivery, and Partial Recharges).
System Fluidity and Queue Mitigation
When scaled from a single vehicle to an entire fleet—such as Autonomous Electric Mobility On-Demand (AEMoD) systems or drone delivery swarms—the benefits of partial charging compound. If every vehicle in a fleet insists on fully recharging, charging infrastructure quickly becomes congested, leading to cascading wait times.
Nasehi et al. demonstrated that proactive route planning that incorporates multiple partial recharges significantly reduces total travel times across an entire network by avoiding congested charging stations (Proactive Route Planning for Electric Vehicles). Instead of viewing charging as an isolated vehicle need, it becomes a tool for network load balancing.
In autonomous ride-hailing fleets, the authors in A Multi-Class Dispatching and Charging Scheme for Autonomous Electric Mobility On-Demand designed partial charging cycles tailored specifically to the energy requirements of distinct trip distances. By matching a vehicle's SoC to the immediate needs of a prioritized customer class, the fleet maintains high availability and reduces customer response times by over 20% (A Multi-Class Dispatching and Charging Scheme for Autonomous Electric Mobility On-Demand). Mathematically, optimizing the ratio of partial to full charging sessions allows a fleet to maintain service stability with significantly fewer total vehicles (Optimal Vehicle Dimensioning for Multi-Class Autonomous Electric Mobility On-Demand Systems).
This cooperative efficiency is also observed in aerial networks. Alkouz et al. showed that when drones in a swarm cooperate by charging only enough to reach their next waypoint rather than fully replenishing, they drastically reduce collective waiting and charging times, which is vital when infrastructure is constrained (Swarm-based Drone-as-a-Service (SDaaS) for Delivery).
Spatial-Temporal Synchronization
Partial charging also serves as a mechanism to synchronize a vehicle's energy state with its spatial-temporal environment. In mobile networks where the chargers themselves are moving, or where mission requirements dictate strict schedules, the amount of energy delivered acts as a clock tuning device.
For instance, in rechargeable wireless sensor networks, He et al. introduced the concept of energy synchronization via the ESync protocol (ESync: an energy synchronized charging protocol for rechargeable wireless sensor networks). Instead of fully charging a node, a mobile charger delivers a precisely calculated partial charge so that the node's next depletion event aligns perfectly with the charger's future proximity on its Traveling Salesman Problem (TSP) tour. This proactive synchronization minimizes node downtime and reduces the travel distance of the charger.
Similarly, in on-demand mobile charging services for EVs, a partial recharging strategy allows service vehicles to distribute power dynamically across multiple stranded or low-battery vehicles without needing to return to a central depot to top off each time (The Mobile Charging Vehicle Routing Problem with Time Windows and Recharging Services).
Boundary Conditions and Strategic Depletion
While partial charging offers clear advantages in time-constrained routing, it operates within a spectrum of energy management strategies. Understanding when not to use it helps clarify why it works.
Long-Term Coverage Tasks
In scenarios where the primary objective is continuous environmental monitoring rather than point-to-point delivery, the math can invert. Meng et al. demonstrated that for multi-agent coverage control tasks, maximizing battery repletion to full capacity is often the most effective strategy (Multi-Agent Coverage Control with Energy Depletion and Repletion). This is because a full charge minimizes the ratio of non-productive travel time (moving back and forth to the charger) to active mission time over long horizons.
Strategic Depletion and Grid Absorption
In hybrid and extended-range electric vehicles (E-REVs), partial charging logic is turned on its head to optimize fuel source transitions. Zhang et al. designed control strategies that aim to deplete the battery to its absolute minimum allowable SoC exactly upon arrival at a charging station (A charging management-based intelligent control strategy for extended-range electric vehicles). By intentionally running the battery down to its lower limit just as it reaches a plug, the system maximizes the vehicle's capacity to absorb clean, cheap grid electricity, thereby minimizing fossil fuel consumption on the final leg of a journey.
This highlights the underlying principle of the entire research direction: energy states should never be managed in isolation. Whether a vehicle is partially charging to skip a queue or strategically depleting to absorb grid power, success relies on tightly coupling the battery's state-of-charge with downstream spatial, temporal, and operational demands.
Key Foundational Work
ESync: an energy synchronized charging protocol for rechargeable wireless sensor networks (2014) — contributed 4 concept elements to this direction.
A charging management-based intelligent control strategy for extended-range electric vehicles (2016) — contributed 2 concept elements to this direction.
A generalized multi-depot vehicle routing problem with replenishment based on LocalSolver (2014) — contributed 1 concept elements to this direction.