Researchers propose multi-agent reinforcement learning framework for robot teams to optimize monitoring accuracy in indoor environments

A learning-based approach uses multi-agent reinforcement learning (MARL — a technique where multiple AI agents learn cooperative behaviors from decentralized observations) to enable robot teams to adjust their motion and directly optimize monitoring accuracy for human activity in indoor spaces.

The framework handles variable numbers of humans and temporal dependencies, and simulation results show the approach outperforms classical coverage, persistent monitoring, and learning-free multi-robot baselines while remaining robust to changes in the number of observed humans.

The work formulates cooperative active observation as a decentralized control problem, addressing applications such as facility management, safety assessment, and space utilization analysis.