Crimson AI NewsA CrimsonLingua Network service
EN ع
← Back to news
Research paper Hugging Face

Adaptive Bridge isolates ROS 2 subscribers from DDS backpressure

AI By Crimson AI Hugging Face Papers 12 September 2026 · 00:00 31 views
Share: X Telegram

A new proxy layer splits ROS 2 topics into separate reliable and best-effort DDS writers to stop one degraded subscriber from stalling safety-critical ones, cutting tail latency from up to 15 seconds to 1.55 ms in the authors' test harness.

Adaptive Bridge isolates ROS 2 subscribers from DDS backpressure

Key points

Researchers have proposed Adaptive Bridge, a proxy-based decoupling layer aimed at a long-standing failure mode in ROS 2 systems that rely on the Data Distribution Service (DDS) for transport. According to the paper, a single network-impaired or throttled subscriber on a RELIABLE topic can trigger backpressure that degrades throughput and latency for every other subscriber sharing the same publisher — including safety-critical nodes — because the publisher's DDS writer can no longer accept new samples.

The proposed bridge sits between publishers and subscribers as a middleman. It subscribes to the original topic and republishes messages to two independent DDS writers: a RELIABLE writer serving critical nodes and a BEST EFFORT writer serving noncritical or degraded nodes. This topic splitting is intended to isolate degraded subscribers so they cannot stall the publisher or the critical path.

Adaptive Bridge also includes a probe-based classifier that monitors subscriber health through sampling with hysteresis and adjusts subscriber rate limits in real time, adding dynamic rate control on top of the static split.

The authors evaluated the system under a Gilbert-Elliott bursty wireless loss model using a reproducible Docker-based harness. They report that the bridge reduced critical-subscriber tail p95 latency from up to 15 seconds to 1.55 ms across all impairment severities tested, while preserving the publisher's configured throughput.

The work is accompanied by a public GitHub repository. The paper's source text does not report additional benchmarks, hardware details, or comparisons against alternative mitigation strategies.

MetricBefore Adaptive BridgeWith Adaptive Bridge
Critical subscriber tail p95 latencyUp to 15 s1.55 ms
Publisher throughputConfigured levelPreserved
Loss modelGilbert-Elliott bursty wirelessGilbert-Elliott bursty wireless
Source
Hugging Face · Hugging Face Papers
Related news
Research paper
Hugging Face 15 Oct 2026

New Survey Maps How Multimodal AI Agents Remember and Evolve

A Hugging Face survey proposes a two-axis framework for multimodal agent memory, separating how information is formed and represen...

6
Research paper
Hugging Face 15 Oct 2026

CriticOPD Fixes On-Policy Distillation by Letting Teachers Critique Student Errors

Hugging Face researchers introduce CriticOPD, a method that restores the teacher's ability to recover from student mistakes during...

7
Research paper
Hugging Face 15 Oct 2026

ActionOPD: Distilling Agent Decisions, Not Tokens

A new Hugging Face paper proposes ActionOPD, an action-level on-policy distillation method that aligns a student agent's decisions...

8