Comprehensive Survey on ROS 2

[Anis_Koubaa]

𝗠𝗶𝗹𝗲𝘀𝘁𝗼𝗻𝗲 𝗔𝗰𝗵𝗶𝗲𝘃𝗲𝗺𝗲𝗻𝘁: 𝗧𝗵𝗲 𝗠𝗼𝘀𝘁 𝗖𝗼𝗺𝗽𝗿𝗲𝗵𝗲𝗻𝘀𝗶𝘃𝗲 𝗦𝘂𝗿𝘃𝗲𝘆 𝗼𝗻 𝗥𝗢𝗦 𝟮:rocket:

I am excited to announce that our latest survey paper,

𝗥𝗢𝗦 𝟮 𝗶𝗻 𝗮 𝗡𝘂𝘁𝘀𝗵𝗲𝗹𝗹: 𝗔 𝗦𝘂𝗿𝘃𝗲𝘆

co-authored with Abdulrahman S. Al-Batati and Dr. Mohamed AbdelKader, is now available on Preprints.org!

Credits go to Abdulrahman S. Al-Batati for the great efforts in gathering this volume of related works and also in building the first repository of ROS/ROS2 publications available at:

𝗥𝗢𝗦/𝗥𝗢𝗦𝟮 𝗥𝗲𝗽𝗼𝘀𝗶𝘁𝗼𝗿𝘆: https://ros.riotu-lab.org/
𝗙𝘂𝗹𝗹 𝗣𝗮𝗽𝗲𝗿: LinkedIn

This study stands as the 𝗺𝗼𝘀𝘁 𝗰𝗼𝗺𝗽𝗿𝗲𝗵𝗲𝗻𝘀𝗶𝘃𝗲 𝘀𝘂𝗿𝘃𝗲𝘆 to date on the transition from 𝗥𝗢𝗦 𝟭 𝘁𝗼 𝗥𝗢𝗦 𝟮, offering a deep dive into the enhancements, challenges, and future directions for ROS 2.

Our analysis covers:
Real-time capabilities
Enhanced modularity
Security improvements
Middleware and distributed systems
Multi-robot system applications

We carefully analyzed 𝟳,𝟰𝟵𝟴 𝗥𝗢𝗦-𝗿𝗲𝗹𝗮𝘁𝗲𝗱 𝗮𝗿𝘁𝗶𝗰𝗹𝗲𝘀, with a focused review of 𝟰𝟯𝟭 𝗥𝗢𝗦 𝟮-𝘀𝗽𝗲𝗰𝗶𝗳𝗶𝗰 𝗽𝘂𝗯𝗹𝗶𝗰𝗮𝘁𝗶𝗼𝗻𝘀, making this a key resource for researchers, developers, and enthusiasts in the ROS community.

Our goal is to provide a cohesive synthesis that helps deepen the understanding of ROS 2’s contributions and guides future research in robotic systems design.

Join us in exploring the potential of ROS 2 and shaping the future of robotics!

hashtag#ROS2 hashtag#Robotics hashtag#ROS hashtag#Research hashtag#AI hashtag#MachineLearning hashtag#RoboticsCommunity hashtag#ROSCommunity hashtag#OpenSource hashtag#Survey

[JRTG]

Congrats on the paper.

A point of criticism though, given you specifically mention to analyse the ‘realtime capabilities’:

For someone new to ROS 2 / DDS / etc., sentences such as the following invoke the expectation that ROS 2 is realtime capable:

With its real-time capabilities, supported by the Data Distribution Service (DDS) middleware, ROS 2 greatly improves communication reliability
and timing for multi-robot systems.

To my knowledge however, current state of ROS 2 is that it simply cannot be considered realtime compliant. There is no bounded-latency guarantee for standard node functionality such as messages and service calls.

Two examples:

• Wrt. services: unless I am mistaken, a service callback allocates heap memory here, which is not realtime compliant.

• Wrt. the mutices used throughout the code: there is currently no support for priority inheritance. A tremendous amount of work has been done wrt. avoiding priority inversion and adding thread attribute configuration, but to my knowledge that work has stalled…

So if you need a true realtime loop “in” ROS 2, my conclusion is that you rather need to implement one in ros2_control, or maybe use e.g. cactus-rt or Orocos (seems no longer maintained). Neither of those options is truly “using ROS 2 for realtime control”, but rather “using ROS 2 for the non-realtime parts around a realtime control loop”.

So I think it is misleading to state that ROS 2 has ‘realtime capabilities’. Either something is realtime, or it is not.

[ZhenshengLee]

https://ros.riotu-lab.org/ the site makes my browser edge no response.

[Anis_Koubaa]

Thank you for your detailed technical feedback. I appreciate this perspective, especially I also originate from the real-time systems community.

Having followed ROS2’s development since its inception in 2014, I’d like to provide some historical and architectural context. A fundamental design distinction between ROS1 and ROS2 was the latter’s focus on real-time support.
While ROS1 operated purely on a best-effort basis using ROSTCP/ROSUDP protocols, ROS2 was architecturally designed to support real-time operations through:

1. The adoption of DDS middleware, which introduced:

• Configurable QoS profiles
• More deterministic communication patterns
• Better support for distributed systems
  2. Real-time oriented design principles, as documented in:
• Understanding real-time programming — ROS 2 Documentation: Foxy documentation
• Introduction to Real-time Systems

You raise valid technical points about current implementation challenges:

• Heap allocations in service callbacks
• Mutex priority inheritance limitations
• The need for frameworks like ros2_control for strict real-time guarantees

Perhaps it would be more precise that we say that ROS2 provides “improved deterministic behavior and real-time support features” compared to ROS1, rather than claiming full real-time capabilities. We will address this change, and thanks for pointing this out.

Also, I think that while the current implementation may partially deviate from the original design goals in terms of strict real-time guarantees, it represents a significant architectural improvement over ROS1’s best-effort approach.

The framework continues to evolve, and while full real-time guarantees might require careful implementation through specialized components like ros2_control or external frameworks, ROS2’s foundational architecture provides better support for real-time requirements than its predecessor.

Thanks again for the feedback

[Anis_Koubaa]

I tried it on different browsers namely Safari, Chrome and FireFox and did not notice any issue. It is a simple code but it may take time to load the whole database and viz it. Can you try other browers?

[Dony]

try to use VPN with different servers, if you are in China. its working

[damulaval]

Nice survey of the ROS packages available out there.

Just a small comment on Table 9: for some reasons, KISS-ICP was categorized as a cyber security package. Based on your categorization, it could be added to the state estimation category or you could also add a category for SLAM, as there are lots of SLAM research papers that use ROS to process sensor data.

Otherwise, if you’re searching for other openly available papers with open-source ROS packages, I think you could find some in the Field Robotics journal.

https://fieldrobotics.net/Field_Robotics/All_Papers.html