Pith. sign in

REVIEW 18 cited by

The Surprising Effectiveness of PPO in Cooperative, Multi-Agent Games

Not yet reviewed by Pith; the record is open.

This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.

SPECIMEN: schema-true, not a live event

T0 review · schema-true

One-sentence machine reading of the paper's core claim.

pith:XXXXXXXX · record.json · timestamp

arxiv 2103.01955 v4 pith:I65B5MFL submitted 2021-03-02 cs.LG cs.AIcs.MA

classification cs.LGcs.AIcs.MA
keywords multi-agentcooperativelearningmethodsoff-policyperformancealgorithmschallenge
verification ladder T0 review T1 audit T2 compute T3 formal
0 comments
read the original abstract

Proximal Policy Optimization (PPO) is a ubiquitous on-policy reinforcement learning algorithm but is significantly less utilized than off-policy learning algorithms in multi-agent settings. This is often due to the belief that PPO is significantly less sample efficient than off-policy methods in multi-agent systems. In this work, we carefully study the performance of PPO in cooperative multi-agent settings. We show that PPO-based multi-agent algorithms achieve surprisingly strong performance in four popular multi-agent testbeds: the particle-world environments, the StarCraft multi-agent challenge, Google Research Football, and the Hanabi challenge, with minimal hyperparameter tuning and without any domain-specific algorithmic modifications or architectures. Importantly, compared to competitive off-policy methods, PPO often achieves competitive or superior results in both final returns and sample efficiency. Finally, through ablation studies, we analyze implementation and hyperparameter factors that are critical to PPO's empirical performance, and give concrete practical suggestions regarding these factors. Our results show that when using these practices, simple PPO-based methods can be a strong baseline in cooperative multi-agent reinforcement learning. Source code is released at \url{https://github.com/marlbenchmark/on-policy}.

Discussion (0). Sign in to comment.

Forward citations

Cited by 18 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score.

  1. Multi-Agent Reinforcement Learning for SLA-Aware Network Slicing in UAV-Enabled MEC

    cs.NI 2026-07 conditional novelty 6.0 of 10

    A predictive MAPPO framework with an SLA-aware reward improves slice SLA stability in UAV-enabled MEC, approaching an informed oracle in simulations.

  2. Offline Nash Solvers Meet Online Tree Search in Multi-Agent Games on Graphs

    cs.GT 2026-07 conditional novelty 6.0 of 10

    Primitive-Guided Tree Search combines offline exact Nash solutions of 1v1/2v1 subgames with online SM-MCTS to produce coordinated multi-agent pursuit policies that outperform PSRO and neural MCTS baselines.

  3. ASALT: Adaptive State Alignment for Lateral Transfer in Multi-agent Reinforcement Learning

    cs.AI 2026-06 unverdicted novelty 6.0 of 10

    ASALT uses observation-level and state-level adapters to align mismatched dimensionalities into a shared embedding for transferring actors and critics in MARL, showing improved sample efficiency and reduced negative t...

  4. A Distributionally Robust Reinforcement Learning Framework for Constrained Urban EV Dispatch

    cs.AI 2026-04 unverdicted novelty 6.0 of 10

    A robust semi-Markov RL agent with MILP feasibility projection and Wasserstein ambiguity set achieves $1.22M net profit on an NYC EV simulator with zero feeder violations, outperforming heuristic and other RL baselines.

  5. Multi-Agent Reinforcement Learning for V2X Resource Allocation: Disentangling MARL Challenges Through Benchmarking

    cs.MA 2026-02 conditional novelty 6.0 of 10

    Robustness to varied vehicle topologies, not partial observability or coordination, is the dominant obstacle for MARL in C-V2X spectrum sharing.

  6. Towards Microgrid Resilience Enhancement via Mobile Power Sources and Repair Crews: A Multi-Agent Reinforcement Learning Approach

    eess.SY 2025-07 conditional novelty 6.0 of 10

    A hierarchical multi-agent reinforcement learning method coordinates mobile power sources and repair crews to restore loads after extreme events in coupled power-transport networks.

  7. Is Inter-Seed Cross-Play Enough? Evaluating the Robustness of Zero-Shot Coordination Algorithms to Implementation Details

    cs.AI 2026-08 conditional novelty 5.0 of 10

    For Other-Play in Yokai, agents trained with different implementation details coordinate across implementations about as well as across seeds, supporting inter-seed cross-play as a proxy for cross-implementation evaluation.

  8. COLMAR: Cooperative View Policy Learning for Multi-Agent Active 3D Reconstruction

    cs.RO 2026-07 conditional novelty 5.0 of 10

    A shared PPO policy with overlap-aware rewards improves multi-agent active 3D reconstruction coverage and accuracy in simulated indoor scenes.

  9. Multi-Agent Reinforcement Learning for SLA-Aware Network Slicing in UAV-Enabled MEC

    cs.NI 2026-07 conditional novelty 5.0 of 10

    Predictive MAPPO with D3QL mobility forecasts reduces slice SLA violation probability and duration in UAV-MEC simulations versus greedy, GA, and random baselines, nearing an informed oracle when forecasts are accurate.

  10. Heterogeneous Policy Networks for Composite Robot Team Communication and Coordination

    cs.RO 2026-06 unverdicted novelty 5.0 of 10

    HetNet achieves 5.84% to 707.65% performance gains and 200x bandwidth reduction over baselines in heterogeneous multi-agent robot teams via graph-attention networks and binarized messaging.

  11. Shape Formation for the Cooperative Transportation of Arbitrary Objects Using Multi-Agent Reinforcement Learning

    cs.RO 2026-06 unverdicted novelty 5.0 of 10

    Multi-agent RL produces robot policies that form balanced supporting shapes under arbitrary objects for cooperative transport and generalize across cluttered scenes and complex geometries.

  12. Enhancing the MADDPG Algorithm for Multi-Agent Learning via Action Inference and Importance Sampling

    cs.LG 2026-06 unverdicted novelty 5.0 of 10

    Action inference and geometric importance sampling enhance MADDPG, yielding better stability, cooperation, and exploration efficiency on the discrete Predator-Prey benchmark.

  13. Decoupled Delay Compensation: Enhancing Pre-trained MARL Policies via Learned Dynamics Filtering

    cs.MA 2026-05 unverdicted novelty 5.0 of 10

    A decoupled estimator combining gated dynamics learning and recursive Kalman filtering improves robustness of pre-trained MARL policies under stale observations and message loss.

  14. Coordination Architecture Shapes Continuous Demand Response Outcomes in Building Districts

    eess.SY 2026-05 unverdicted novelty 5.0 of 10

    In a 25-building district simulation, the hybrid MPC-SAC architecture delivered the strongest balance of load tracking accuracy (4.8% NMBE), thermal comfort (16.8% exceedance), and lowest spatial variability compared ...

  15. A Distributionally Robust Reinforcement Learning Framework for Constrained Urban EV Dispatch

    cs.AI 2026-04 unverdicted novelty 5.0 of 10

    PD-RSAC, a distributionally robust SAC variant with GCN encoder and MILP constraint projection, reports $1.22M net profit on an NYC taxi-based EV simulator while achieving zero feeder violations, outperforming heurist...

  16. Self-Supervised Goal-Reaching Results in Multi-Agent Cooperation and Exploration

    cs.LG 2025-09 conditional novelty 5.0 of 10

    Self-supervised multi-agent goal-reaching, where each agent independently learns a contrastive critic of its own observations, achieves cooperation and exploration in sparse-reward MARL tasks where standard baselines fail.

  17. $\alpha$-fair heterogeneous agent reinforcement learning

    cs.MA 2026-06 unverdicted novelty 4.0 of 10

    Introduces α-fair HATRPO and HAPPO algorithms that integrate α-fairness into HATRL via a weighted advantage function while claiming to preserve convergence to Nash equilibria.

  18. Multi-agent Reinforcement Learning for Robotized Coral Reef Sample Collection

    cs.RO 2025-07 conditional novelty 4.0 of 10

    An RL controller for coral sample collection, trained in Unity, is transferred zero-shot to a physical BlueROV2 using real-time underwater motion capture to drive the digital twin.

Pith tools