Scaling MADDPG for Enhanced Multi-Agent Reinforcement Learning

Addressing multi-agent scenarios is a challenging task for traditional reinforcement learning algorithms due to the environments' dynamic nature from each agent's perspective. In this research, we delve into the performance of multi-agent deep deterministic policy gradients (MAD-DPG), a tailored algorithm for multi-agent scenarios, in the context of a cooperative navigation OpenAI environment involving three agents. Through extensive experimentation, we investigate the impact of various hyperparameters and action exploration strategies on MADDPG's performance. Moreover, we propose enhancements to MAD-DPG to overcome its main weakness, which lies in its limited scalability to more significant numbers of agents. Our findings demonstrate that MAD-DPG outperforms singleagent policy gradient approaches significantly. Furthermore, our innovative improvements to MAD-DPG effectively tackle its primary limitation, allowing the algorithm to scale more efficiently to accommodate larger numbers of agents while only minimally impacting its overall performance.

• Publication year

  2025

• Venue

  2025 Second International Conference on Pioneering Developments in Computer Science & Digital Technologies (IC2SDT)

• Publication date

  2025-12-04

• Fields of study

  Not labeled

• Identifiers
  DOI 10.1109/IC2SDT68218.2025.11383636
• External record

  Open on Semantic Scholar

• Source metadata

  Semantic Scholar

• No claims are published for this paper.

• No concepts are published for this paper.

• Graph Neural Network Meets Multi-Agent Reinforcement Learning: Fundamentals, Applications, and Future Directions

  2024cited by this paper
• Multiagent Deep Deterministic Policy Gradient-Based Computation Offloading and Resource Allocation for ISAC-Aided 6G V2X Networks

  2024cited by this paper
• Deep Hierarchical Communication Graph in Multi-Agent Reinforcement Learning

  2023cited by this paper
• Scaling laws for single-agent reinforcement learning

  2023cited by this paper
• Broken Neural Scaling Laws

  2022cited by this paper
• Learning structured communication for multi-agent reinforcement learning

  2020cited by this paper
• Counterfactual Multi-Agent Policy Gradients

  2017cited by this paper
• Parameter Space Noise for Exploration

  2017cited by this paper
• Intrinsic Motivation and Automatic Curricula via Asymmetric Self-Play

  2017cited by this paper
• Multiagent Bidirectionally-Coordinated Nets for Learning to Play StarCraft Combat Games

  2017cited by this paper
• Multi-Agent Actor-Critic for Mixed Cooperative-Competitive Environments

  2017influential reference
• Learning to Communicate with Deep Multi-Agent Reinforcement Learning

  2016cited by this paper
• Continuous control with deep reinforcement learning

  2015cited by this paper
• Coordinated Multi-Robot Exploration Under Communication Constraints Using Decentralized Markov Decision Processes

  2012cited by this paper
• Independent reinforcement learners in cooperative Markov games: a survey regarding coordination problems

  2012cited by this paper
• Multi-agent Reinforcement Learning: An Overview

  2010cited by this paper
• The Ornstein–Uhlenbeck process as a model of a low pass filtered white noise

  2008cited by this paper
• A Comprehensive Survey of Multiagent Reinforcement Learning

  2008cited by this paper
• Shaping multi-agent systems with gradient reinforcement learning

  2007cited by this paper
• Coordination in multiagent reinforcement learning: a Bayesian approach

  2003cited by this paper
• The Dynamics of Reinforcement Learning in Cooperative Multiagent Systems

  1998cited by this paper
• Planning, Learning and Coordination in Multiagent Decision Processes

  1996cited by this paper
• Multiagent reinforcement learning in the Iterated Prisoner's Dilemma.

  1996cited by this paper

• No citing papers are available for this paper.