Route Recommendations for Traffic Management Under Learned Partial Driver Compliance

TL;DR

This work tackles the gap between system-optimal traffic assignment and partial driver adherence by formalizing a compliance-aware routing framework. It computes a system-optimal edge flow $x_e^*$ via flow optimization, learns a compliance probability model $\hat{\phi}$ from historical data, and solves a stochastic route-allocation IP to minimize the occupancy gap $|L_e^*-\sum_n\sum_p \phi_n(p|p_n^r)\mathbb{I}(e\in p)|$ with $L_e^*=x_e^* t_e(x_e^*)$. The contributions include a compliance-aware formulation, a data-driven compliance learner with Random Forest achieving substantial predictive accuracy, and an IP-based route allocation that integrates learned compliance to closely approximate system-optimal performance. Numerical experiments on a grid network demonstrate travel-time improvements and the practicality of incorporating learned driver behavior into congestion management, with potential extensions to incentives and multi-modal transportation.

Abstract

In this paper, we aim to mitigate congestion in traffic management systems by guiding travelers along system-optimal (SO) routes. However, we recognize that most theoretical approaches assume perfect driver compliance, which often does not reflect reality, as drivers tend to deviate from recommendations to fulfill their personal objectives. Therefore, we propose a route recommendation framework that explicitly learns partial driver compliance and optimizes traffic flow under realistic adherence. We first compute an SO edge flow through flow optimization techniques. Next, we train a compliance model based on historical driver decisions to capture individual responses to our recommendations. Finally, we formulate a stochastic optimization problem that minimizes the gap between the target SO flow and the realized flow under conditions of imperfect adherence. Our simulations conducted on a grid network reveal that our approach significantly reduces travel time compared to baseline strategies, demonstrating the practical advantage of incorporating learned compliance into traffic management.

Figures (5)

• Figure 1: System-optimal route recommendation framework under partial driver compliance.
• Figure 2: Example $4\times4$ grid traffic network and its initial (base) flow distribution. Nodes marked as red diamonds serve as designated origins and destinations. Each edge label indicates the base flow rate.
• Figure 3: Comparison of the model's predicted compliance probabilities against actual observations, illustrating the strong alignment of predictions near the diagonal and indicating robust predictive accuracy.
• Figure 4: Confusion matrix for the compliance prediction model. The dominance of diagonal entries underscores that the model is reliably predicting compliance behavior.
• Figure 5: Comparison of different route recommendation strategies under different levels of compliance information. The left panel shows the sum of deviation from the system-optimal flow in all edges, while the right panel illustrates total travel time. Lower values indicate better performance.