Increased Plan Stability in Cooperative Electric Vehicles Path-Planning
Keywords: Electric Vehicles, Multi-Agent Planning, Plan Stability, Human-Aware Planning, Predictable Behavior
TL;DR: The paper presents an extension to the Cooperative Electric Vehicles Path-Planning (CEVPP) problem, which focuses on enhancing plan stability to alleviate cognitive load on human drivers.
Abstract: The Cooperative Electric Vehicles Planning Problem (CEVPP) has recently been proposed as a multi-agent variant of the Electric Vehicle Path-Planning Problem (EVPP). It consists in finding a set of paths for a fleet of electric vehicles that minimizes the global plan execution time, including the time spent waiting at the charging stations. In the proposed formulation, new Electric Vehicles (EVs) can join the fleet at any time, and a centralized planner recomputes the optimal plan every now and then to take them into account. However, the newly computed plans of EVs that were already on the road can change drastically, compared to their previous plans. In this paper, we propose an extension of CEVPP that considers the plan stability in the objective function as a way to reduce cognitive load on the human drivers. The results of our experiments, conducted with real road networks and charging stations, indicate that our approach can significantly reduce the variability of the optimal plans, while keeping low the global plan execution time.
Submission Number: 7
Loading