COMPASS: Cooperative Multi-Agent Persistent Monitoring using Spatio-Temporal Attention Network

2025-07-22

AI Summary

Persistent monitoring of multiple moving targets by multi-agent systems in dynamic environments poses significant challenges in coordination, adaptability, and uncertainty handling. Method: This paper proposes a decentralized collaborative framework integrating graph neural networks (GNNs) with spatiotemporal attention mechanisms and Gaussian process (GP) modeling to capture epistemic uncertainty. The framework enables history-aware perception, spatial-contextual reasoning, and uncertainty-driven re-visitation planning. It adopts a centralized-training-with-decentralized-execution paradigm, optimizing cooperative policies via multi-agent reinforcement learning. Contribution/Results: Compared to state-of-the-art baselines, the method achieves substantial improvements in target coverage ratio, regional revisit efficiency, and uncertainty suppression. Experimental results validate the effectiveness and robustness of jointly leveraging spatiotemporal attention and probabilistic uncertainty modeling for dynamic persistent monitoring tasks.

Technology Category

• Multiagent Systems: Multiagent Systems under Uncertainty
• Reasoning under Uncertainty: Sequential Decision Making
• Intelligent Robots: Multimodal Perception & Sensor Fusion

Application Category

• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• User Modeling, Personalization and Recommendation: Attacks and countermeasures in recommendation systems

Abstract

Persistent monitoring of dynamic targets is essential in real-world applications such as disaster response, environmental sensing, and wildlife conservation, where mobile agents must continuously gather information under uncertainty. We propose COMPASS, a multi-agent reinforcement learning (MARL) framework that enables decentralized agents to persistently monitor multiple moving targets efficiently. We model the environment as a graph, where nodes represent spatial locations and edges capture topological proximity, allowing agents to reason over structured layouts and revisit informative regions as needed. Each agent independently selects actions based on a shared spatio-temporal attention network that we design to integrate historical observations and spatial context. We model target dynamics using Gaussian Processes (GPs), which support principled belief updates and enable uncertainty-aware planning. We train COMPASS using centralized value estimation and decentralized policy execution under an adaptive reward setting. Our extensive experiments demonstrate that COMPASS consistently outperforms strong baselines in uncertainty reduction, target coverage, and coordination efficiency across dynamic multi-target scenarios.

Problem

• Persistent monitoring of dynamic targets using multi-agent reinforcement learning
• Decentralized agents efficiently track multiple moving targets under uncertainty
• Spatio-temporal attention network integrates historical and spatial context for planning

Innovation

• Decentralized MARL framework for persistent monitoring
• Spatio-temporal attention network integrates historical data
• Gaussian Processes model target dynamics uncertainty