Combinatorial Optimization Augmented Machine Learning for Dynamic Electric Autonomous Dial-a-Ride Problem

This paper addresses the Dynamic Electric Autonomous Dial-a-Ride Problem (Dyn-EADARP) by proposing a decision-epoch-based framework. At each decision epoch, the system determines which passenger requests to serve immediately versus postpone, while simultaneously planning vehicle routes, schedules, and charging operations. The approach combines combinatorial optimization with machine learning to handle the dynamic nature of incoming requests in electric autonomous vehicle fleets.