Measurement of Regional Electric Vehicle Adoption Using Multiagent Deep Reinforcement Learning

Abstract:This study investigates socioeconomic disparities in the early adoption of Electric Vehicles (EVs) in the United States and proposes a solution using a multiagent deep reinforcement learning-based policy simulator. Testing this model with data from Austin, Texas, reveals that neighborhoods with higher incomes and predominantly White demographics lead in EV adoption. To address disparities, tiered subsidies were introduced, with increasing amounts for low-income communities. Results show that narrowing the adoption gap began when incentives increased from 20% to 30% . This framework offers a novel approach to testing policy scenarios for promoting EV adoption, with potential for further development and expansion in future studies.