Stay, Reposition, or Recharge: Battery-Aware Operation of Robotaxi Fleets
stat.AP, econ.GN, q-fin.EC
Submitted: 2026-11-05
Updated: 2026-11-05
License: http://creativecommons.org/licenses/by/4.0/
The gist: Electric robotaxis are rapidly expanding from pilots to scaled commercial deployment; unlike traditional human-driven ride-hailing vehicles…
Terminology
Abstract
Electric robotaxis are rapidly expanding from pilots to scaled commercial deployment; unlike traditional human-driven ride-hailing vehicles, they require centralized fleet control and careful management of vehicle state of charge (SoC). We study how a fleet manager should jointly make service, empty repositioning, idling, and charging decisions in such systems. We develop a continuous-time fluid model in which vehicles evolve across location-charge states, and identify two forms of charge-aware control that are absent from classic repositioning models: early return-to-depot (RTD) and charge-aware repositioning. In particular, the optimal solution admits simple threshold structures. Each location has a battery threshold such that vehicles should return to the depot to recharge when their SoC is below it and should not do so when their SoC is above it. Vehicles should not reposition below the same threshold either; above it, the choice of destination depends on whether a vehicle's SoC matches the destination's battery sensitivity, a notion we introduce in the paper. We then quantify the value of charge awareness by comparing fully charge-aware policies with charge-oblivious policies that ignore SoC in RTD and repositioning decisions. We show that the performance gap can be unbounded, driven by a dilemma between under-utilizing high-charge vehicles and over-sending low-charge vehicles to locations from which charging is costly. A numerical case study calibrated using 2025 NYC Taxi and Limousine Commission trip records illustrates these thresholds and shows that charge-aware control can generate substantial gains in a realistic urban network. Importantly, simple threshold-based heuristics recover most of the value of the fully charge-aware solution.
Related papers
- Unraveling time-varying causal effects of multiple exposures: integrating Functional Data Analysis with Multivariable Mendelian Randomization
- Strategic Causal Policy Learning with Manipulable Eligibility
- BayesJudge: Uncertainty-Aware Bayesian Meta-Evaluation of Human and LLM Judgments
- Automating structural reliability analysis with a multi-agent large language model framework
- Improving Minority Population Sampling with BISG Probabilities: Evidence from a Survey of Jewish Americans
- When Do Sports Leagues Stop Being Competitive? Bayesian Detection of Competitive-Balance Eras from Match-Level Data