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

Name: Robust Multi-Agent Reinforcement Learning for Small UAS Separation Assurance under GPS Degradation and Spoofing
Uploaded: 2026-09-02T19:25:05.174347+00:00
Duration: 396 s
Description: The episode discusses a paper on Robust Multi-Agent Reinforcement Learning for small drones (UAS) to ensure safe separation despite GPS degradation or spoofing. The system uses MARL and specific regularization techniques to maintain safety. Simulation results demonstrate that even with 35% corrupted GPS data, the robust policy maintains near-zero collisions.

The episode discusses a paper on Robust Multi-Agent Reinforcement Learning for small drones (UAS) to ensure safe separation despite GPS degradation or spoofing. The system uses MARL and specific regularization techniques to maintain safety. Simulation results demonstrate that even with 35% corrupted GPS data, the robust policy maintains near-zero collisions.

Open the full episode