ReDAG-RT: Global Rate-Priority Scheduling for Real-Time Multi-DAG Execution in ROS 2
New user-space scheduler for ROS 2 reduces 99th percentile response times by 42.9% for real-time robots.
A team of researchers has unveiled ReDAG-RT, a novel scheduling framework designed to solve a critical bottleneck in modern robotics software. The Robot Operating System 2 (ROS 2) is the dominant middleware for building robotic applications, where functions like perception and control are structured as directed acyclic graphs (DAGs) of callbacks. The default executors in ROS 2 use a best-effort approach, which can lead to callback contention, priority inversion, and missed deadlines when multiple complex DAGs run simultaneously. This unpredictability has been a major barrier for deploying ROS 2 in safety-critical, real-time systems like self-driving cars or surgical robots.
ReDAG-RT addresses this by implementing a global, user-space scheduler that operates on top of the unmodified ROS 2 core. Its key innovation is a Rate-Priority driven ready queue that orders callback execution based on their activation rates, a principle from classical Rate-Monotonic scheduling theory. The framework also enforces concurrency bounds per DAG to prevent any single graph from monopolizing resources. In experiments using the ROS 2 Humble distribution, ReDAG-RT demonstrated a 29.7% reduction in deadline misses and a 42.9% improvement in worst-case (99th percentile) response times compared to standard executors.
The significance lies in its practical implementation. By achieving deterministic scheduling entirely in user-space, ReDAG-RT provides a path to certifiable, real-time performance for complex robotic systems without requiring changes to the ROS 2 API, its executors, or the underlying operating system scheduler. This work bridges the gap between academic real-time systems theory and industrial robotic software development, offering a deployable solution for engineers building the next generation of autonomous machines.
- Cuts deadline miss rates by 29.7% and improves 99th percentile response times by 42.9% vs. standard ROS 2 executors.
- Implements a global Rate-Priority scheduler in user-space, requiring no modifications to the core ROS 2 API or OS.
- Enables deterministic, analyzable execution for safety-critical robotic systems like autonomous vehicles and industrial automation.
Why It Matters
Enables reliable, predictable robotics for autonomous vehicles and industrial systems, moving ROS 2 closer to safety-critical certification.