Title : An explainable multi-agent AI framework for sustainable autonomous mobility and electric grid integration
Abstract:
The rapid expansion of autonomous mobility and electric vehicle ecosystems creates significant opportunities for energy-efficient transportation, while also introducing complex challenges in optimizing their integration with renewable-powered electric grids. This study introduces an agent-based, explainable artificial intelligence framework designed to investigate sustainability-oriented decision-making in autonomous routing, electric vehicle charging, and grid operations. The framework simulates a virtual city environment in which autonomous vehicle agents, charging station agents, and grid operator agents interact through interpretable policies that explicitly consider energy consumption, carbon intensity, and renewable energy availability. Machine learning, reinforcement learning, and post-hoc explainability techniques are integrated to produce transparent, sustainability-aligned decisions in real time. Through comprehensive simulation of multi-agent interactions, the framework serves as a controlled platform for examining the influence of explainability on system performance, sustainability metrics, and user trust. Simulation results facilitate the evaluation of trade-offs between interpretable and black-box models, measurement of the effects of grid-aware routing and charging strategies, and assessment of grid stability under varying mobility and renewable energy scenarios. This research provides a novel and accessible approach for advancing explainable and sustainable artificial intelligence in autonomous mobility and energy systems, supporting broader initiatives to accelerate the transition toward cleaner, more transparent, and more resilient transportation-energy ecosystems.
