SFG-ROS: A Resource-Aware Framework for Dense Multi-Agent Perception

Researchers introduced SFG-ROS, a ROS 2-based framework for dense multi-robot perception that reduces network load and CPU overhead.

It uses schema-driven traffic routing, centralized sensor-data decoding, and hardware-agnostic containers to better support heterogeneous robot fleets.

Tests on wheeled and legged robots showed bounded network traffic and a 72.3% reduction in CPU scaling penalty versus standard ROS 2.

The approach addresses key scalability bottlenecks in collaborative robotics, making mixed-fleet perception pipelines more practical in real deployments.