Including driver reaction time in ride-hailing dispatch: a filtering-grabbing optimisation approach

The rapid expansion of the ride-hailing industry has heightened the demand for dispatch strategies that ensure both fairness in driver-order matching and operational efficiency for platforms. However, prevailing dispatch strategies exhibit inherent limitations. Furthermore, heterogeneity in drivers’ order acceptance times significantly impacts dispatch outcomes but is largely overlooked. To address these deficiencies, this study incorporates driver reaction time into ride-hailing dispatch systems and proposes a hybrid Filtering-Grabbing strategy that synthesises System-Assigning and Driver-Grabbing paradigms. Driver reaction time is decomposed into decision and action phases and modelled using a Markov decision process combined with fuzzy control. A multi-objective optimisation model jointly minimises dispatch distance and order acceptance time, solved via a hill-climbing algorithm. Validated on Didi Gaia data from Chengdu across multiple periods, the sensitivity analysis shows robustness. Compared to benchmark strategies, utility value decreases by an average of 11.53% off-peak and 48.54% peak while balancing acceptance time and dispatch distance.