Deep Reinforcement Learning-Based Path Planning for Dynamic and Heterogeneous Environments

Abstract

This study investigates the application of Deep Reinforcement Learning (DRL) algorithms, namely Double Deep Q-Network (DDQN) and Deep Deterministic Policy Gradient (DDPG), in the context of path planning within dynamic and heterogeneous environments. Recognizing the limitations of traditional path planning methods in such complex settings, this research evaluates the suitability of DRL for more adaptable and dynamic navigation solutions. Through extensive simulations, we compare the performance of DDQN and DDPG in various scenarios, highlighting their respective strengths and limitations. DDQN demonstrates higher efficiency in static environments, while DDPG excels in dynamic conditions due to its adaptability and continuous action capabilities. Our findings contribute significant insights into the application of DRL in robotic navigation, particularly in environments characterized by their unpredictability and diversity. This paper not only furthers the understanding of DRL in complex path planning scenarios but also provides guidance for the practical implementation of these algorithms in real-world autonomous systems.