Learning Selective Communication for Multi-Agent Path Finding

Ma, Ziyuan; Luo, Yadan; Pan, Jia

doi:10.1109/lra.2021.3139145

Cited by 23 publications

(9 citation statements)

References 22 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…This assumption is driven by the aim to enable a more realistic and decentralized deployment in real-world scenarios. Agents' observations are usually categorized as either a bird's-eye-view (BEV) of the agent's surroundings [67], [151], [153], [155]- [157] or a firstperson-view (FPV) based on sensor data [149], [163].…”

Section: ) Simulator Environmentsmentioning

confidence: 99%

A Comprehensive Review on Leveraging Machine Learning for Multi-Agent Path Finding

Alkazzi,

Okumura

2024

IEEE Access

View full text Add to dashboard Cite

This review paper provides an in-depth analysis of the latest advancements in applying Machine Learning (ML) to solve the Multi-Agent Path Finding (MAPF) problem. The MAPF problem is about finding collision-free paths for multiple agents to travel from their source to goal locations in a known environment. This method underpins a range of advanced, large-scale automated systems, notably in warehouse logistics. The existing research on conventional MAPF is extensive; however, recent developments in ML have notably augmented the capabilities of MAPF techniques. This research seeks to thoroughly investigate the emerging field focused on using ML to help solve the MAPF problem. It aims to highlight the transformative potential of ML in enhancing the efficiency and effectiveness of multi-agent systems in navigating and coordinating in complex environments. Our study comprehensively examines the entire MAPF process, encompassing environment representation, path planning, and solution execution.

show abstract

Section: ) Simulator Environmentsmentioning

confidence: 99%

A Comprehensive Review on Leveraging Machine Learning for Multi-Agent Path Finding

Alkazzi,

Okumura

2024

IEEE Access

View full text Add to dashboard Cite

show abstract

“…Communication topologies can be classified into three types: predefined fixed topology [30,39,56,63,[69][70][71], predefined dynamic topology [57,58,60,61,65,72,75] and learned topology [31,40,59,62,64,67]. Predefined fixed topologies are manually defined before training based on prior assumptions and remain fixed throughout training and execution, e.g., global communications (complete graph among agents).…”

Section: Communication Topologymentioning

confidence: 99%

“…Similarly, predefined dynamic topologies rely on a pre-defined, nonlearnable condition to decide whether communication would occur between any given pair of agents at every time step. For example, many works have either assumed a fixed communication range around each agent [57,60,61,65,72], a threshold on the confidence level of each agent's individual decision [58], or the influence of the presence of other agents for decision adjustment [75].…”

Section: Communication Topologymentioning

confidence: 99%

Distributed Reinforcement Learning for Robot Teams: A Review

Wang¹,

Damani²,

Wang³

et al. 2022

Preprint

View full text Add to dashboard Cite

Purpose of review: Recent advances in sensing, actuation, and computation have opened the door to multi-robot systems consisting of hundreds/thousands of robots, with promising applications to automated manufacturing, disaster relief, harvesting, last-mile delivery, port/airport operations, or search and rescue. The community has leveraged model-free multi-agent reinforcement learning (MARL) to devise efficient, scalable controllers for multi-robot systems (MRS). This review aims to provide an analysis of the state-of-the-art in distributed MARL for multi-robot cooperation. Recent findings: Decentralized MRS face fundamental challenges, such as non-stationarity and partial observability. Building upon the "centralized training, decentralized execution" paradigm, recent MARL approaches include independent learning, centralized critic, value decomposition, and communication learning approaches. Cooperative behaviors are demonstrated through AI benchmarks and fundamental real-world robotic capabilities such as multi-robot motion/path planning. Summary: This survey reports the challenges surrounding decentralized model-free MARL for multi-robot cooperation and existing classes of approaches. We present benchmarks and robotic applications along with a discussion on current open avenues for research.

show abstract

“…In recent years, multi-agent reinforcement learning (MARL) has made remarkable strides in enhancing cooperative robot tasks and distributed control domains, exemplified by traffic lights control (Chu, Chinchali, and Katti 2020) and robots navigation (Han, Chen, and Hao 2020). Given the inherent partial observability in multi-agent tasks, several researchers have explored the integration of communication mechanisms to facilitate information exchange among agents (Ma, Luo, and Pan 2021).…”

Section: Introductionmentioning

confidence: 99%

Robust Communicative Multi-Agent Reinforcement Learning with Active Defense

Yu,

Qiu,

Yao

et al. 2024

AAAI

View full text Add to dashboard Cite

Communication in multi-agent reinforcement learning (MARL) has been proven to effectively promote cooperation among agents recently. Since communication in real-world scenarios is vulnerable to noises and adversarial attacks, it is crucial to develop robust communicative MARL technique. However, existing research in this domain has predominantly focused on passive defense strategies, where agents receive all messages equally, making it hard to balance performance and robustness. We propose an active defense strategy, where agents automatically reduce the impact of potentially harmful messages on the final decision. There are two challenges to implement this strategy, that are defining unreliable messages and adjusting the unreliable messages' impact on the final decision properly. To address them, we design an Active Defense Multi-Agent Communication framework (ADMAC), which estimates the reliability of received messages and adjusts their impact on the final decision accordingly with the help of a decomposable decision structure. The superiority of ADMAC over existing methods is validated by experiments in three communication-critical tasks under four types of attacks.

show abstract

Learning Selective Communication for Multi-Agent Path Finding

Cited by 23 publications

References 22 publications

A Comprehensive Review on Leveraging Machine Learning for Multi-Agent Path Finding

A Comprehensive Review on Leveraging Machine Learning for Multi-Agent Path Finding

Distributed Reinforcement Learning for Robot Teams: A Review

Robust Communicative Multi-Agent Reinforcement Learning with Active Defense

Contact Info

Product

Resources

About