Robust Multi-Agent Reinforcement Learning for Small UAS Separation Assurance under GPS Degradation and Spoofing

📰 ArXiv cs.AI

Robust Multi-Agent Reinforcement Learning ensures separation assurance for small UAS under GPS degradation and spoofing

advanced Published 1 Apr 2026

Action Steps

Formulate the problem as a zero-sum game between agents and an adversary
Implement Multi-Agent Reinforcement Learning (MARL) to learn robust policies
Evaluate the performance of MARL under various GPS degradation and spoofing scenarios
Refine the MARL algorithm to improve separation assurance and adapt to changing conditions

Who Needs to Know This

This research benefits teams of ai-engineers, ml-researchers, and software-engineers working on autonomous systems, particularly those involved in drone development and air traffic control, as it provides a robust method for maintaining separation assurance in challenging environments

Key Insight

💡 MARL can effectively mitigate the effects of GPS corruption on air traffic state estimation

Key Takeaways

Robust Multi-Agent Reinforcement Learning ensures separation assurance for small UAS under GPS degradation and spoofing

Full Article

Title: Robust Multi-Agent Reinforcement Learning for Small UAS Separation Assurance under GPS Degradation and Spoofing

Abstract:
arXiv:2603.28900v1 Announce Type: cross Abstract: We address robust separation assurance for small Unmanned Aircraft Systems (sUAS) under GPS degradation and spoofing via Multi-Agent Reinforcement Learning (MARL). In cooperative surveillance, each aircraft (or agent) broadcasts its GPS-derived position; when such position broadcasts are corrupted, the entire observed air traffic state becomes unreliable. We cast this state observation corruption as a zero-sum game between the agents and an adver

Read full paper → ← Back to Reads