Skip to main content

Meetings

This page lists all the meetings that have occurred as part of the working group.

All meetings are public and held every two weeks from 1600-1700 UTC. Please feel free to join us. There are some details below for extra information:

If you have an interesting topic of discussion for the group, please propose that topic using the Github Issue.

2026-06-15: Logging and Observability Discussion (Upcoming)​

The group has been discussing Logging and Observability tips and best practices over the course of several months. In this session, the group will bring together some of their learnings as an informal discussion. Notes from this meeting will serve as a basis for the Logging and Observability Guide; once the guide is written, previous guest speakers will be invited to review it before it's published.

2026-06-01: Part 2 - Preparing for State of Cloud Robotics Survey​

In this session, the group continued to update the set of questions for the next State of Cloud Robotics Survey. They successfully built a set of questions and the survey was sent out the next day. Any developers using cloud robotics are invited to respond to the survey.

If you'd like to see the meeting, is it available on YouTube:

2026-05-18: Preparing for State of Cloud Robotics Survey​

In this session, the group started preparing for a new State of Cloud Robotics Survey by going over the questions sent in the previous survey. The previous survey asked respondents if they would suggest any new questions, and the group used feedback from this question to come up with new questions ready for the next survey.

If you'd like to see the meeting, is it available on YouTube:

2026-05-04: Custom Capabilities in Transitive Robotics - Again​

In this session, the group continued their Transitive Robotics tryout by trying one of the more advanced features: writing and deploying a custom capability. This feature allows customers to write their own custom code and deploy it to their robots alongside the features available directly from Transitive Robotics.

We were able to get a working setup over the majority of the meeting, and by the end had managed to implement a small piece of custom functionality.

If you'd like to see the meeting, it is available on YouTube:

2026-04-20: Custom Capabilities in Transitive Robotics​

In this session, the group attempted to set up a custom capability in Transitive Robotics. Unfortunately, the attempted setup was on a Windows machine, which isn't a supported way to setting up the development environment. Hence, this session has again been postponed until a supported development environment can be used.

This session was recorded, and consists of the host trying to set up the development environment in WSL and on a VirtualBox Virtual Machine, neither of which were successful. If you are interested in seeing the recording, it is available here:

2026-03-23: Transitive Robotics Tryout​

The group used this session to try out Transitive Robotics, which is a service that allows users to deploy and manage robots, including giving full-stack robotic capabilities. Capabilities include data capture and storage, which makes Transitive Robotics a useful case study for our focus on Logging & Observability.

We managed to set up a remote Turtlebot with Transitive Robotics and remote control the robot. We also added Maps as a capability, which unfortunately didn't work due to incompatibility with ROS 2 Jazzy (support has since been added for Jazzy).

If you'd like to see the meeting, it is available on YouTube:

2026-03-09: Part 2: Canonical Observability Stack Tryout​

In this session, the group continued our tryout of the Canonical Observability Stack (COS) from the previous meeting. We were successful in hosting the full stack and viewing the public pages, as well as connecting a simulated robot to the stack. We could view logs and system statistics from the simulated robot.

If you'd like to see the meeting, it is available on YouTube:

2026-02-25: Canonical Observability Stack Tryout​

Following the prior meeting's presentation on the Canonical Observability Stack (COS), the group attempted to deploy COS components based on the publicly available tutorials and documentation. The recording unfortunately cut out shortly into the meeting due to lack of disk space.

From that point, the group worked through setting up an AWS instance to host the COS server side, and a VirtualBox VM to host the robot side. By the end of the session, the group had a slightly misconfigured server, so the robot was unable to connect. Hence, this will be continued in the next session.

If you're interested to watch the recorded part of the meeting, it is available on YouTube:

2026-02-11: Canonical Observability Stack Tryout (Cancelled)​

Following the previous meeting's presentation on the Canonical Observability Stack, the group will attempt to deploy some components based on the publicly available tutorials and documentation. This is similar to previous meetings where the group tried out KubeEdge and Zenoh.

2026-01-28: Canonical Observability Stack with Guillaume Beuzeboc​

For this session, the CRWG invited Guillaume Beuzeboc from Canonical to present on the Canonical Observability Stack (COS). COS is a general observability stack for devices such as drones, robots, and IoT devices. It operates from telemetry data, and the COS team has extended it to support robot-specific use cases. Guillaume, a software engineer at Canonical, previously presented COS at ROSCon 2025 and kindly agreed to join this meeting to discuss additional technical details with the CRWG.

The talk also includes questions from the group members, such as how non-experts can operate the tool, and whether it supports conditional deployment based on robot capabilities.

If you'd like to see the meeting, it is available on YouTube:

2026-01-14: ROSCon Review Continued​

In this meeting, the group reviewed the list of ROSCon talks collected from the previous session. They reviewed the talks most likely to be relevant to Logging & Observability by looking through the extra information and transcripts from the talks. The group also made a few points about the talks that will be summarised in an upcoming blog post.

If you'd like to see the meeting, it is available on YouTube:

2025-12-31: No Meeting​

The group agreed to skip this meeting as it is the holiday period for many members.

2025-12-17: Cancelled Meeting​

The group met in this session, but given the few members available, decided to postpone the meeting until the new year.

2025-12-03: No Session​

The group will be skipping this session as some members are unavailable.

2025-11-19: ROSCon Review​

In this meeting, the group looked at the talks listed for ROSCon, examining the description of each for any that would be relevant to Logging & Observability, the current focus of the group. Next session will dive deeper into the talks to pull out interesting tips for a blog post.

If you'd like to see the meeting, it is available on YouTube:

2025-11-05: General Catch-Up​

The CRWG used this meeting to have a general catch-up. They answered some questions about the use of cloud to view remote video data, and discussed the past few sessions. The group also decided to have a ROSCon review session.

If you'd like to see the meeting, it is available on YouTube:

2025-10-22: INSAION with Sergi Grau-Moya and Victor Massagué Respall​

In this meeting, Sergi Grau-Moya, Co-founder and CEO of INSAION, and Victor Massagué Respall, Co-founder and CTO of INSAION, presented the INSAION platform to the group. This platform is an observability platform for your robot fleet, allowing users to optimise robot operations and explore advanced robot diagnostics. It also has some very impressive visualisations!

The group then had time to ask follow-up questions, including what big takeaways INSAION had from customer conversations, and how to maintain a log of previous resolutions to apply to future issues.

If you'd like to see the meeting, it is available on YouTube:

2025-10-08: AMP with Carter Schultz​

In this meeting, Carter Schultz of AMP came to discuss how the company manages large deployments and the pain points they see from doing so.

This was a fascinating talk and highlighted a lot of issues that AMP had to hand-build solutions for. Carter also spoke about how many of the newer tools available on the market now do not fit well into their architecture, which makes them less enticing to adopt.

Some key takeaways for the group included:

  • Don't put metrics in logs! If you need state, put it on a topic, then egress through the topic. Logs should always serve a purpose.
  • Build alerts with an action that can resolve them. Operations managers will start to ignore red alerts if there's never something they can do to resolve them.

The CRWG had a lot of questions about the talk afterwards, such as whether newer technologies would be useful if AMP was starting up now, and how AMP manages redundant data uploads.

If you'd like to see the meeting, it is available on YouTube:

2025-09-24: ROSBag MCP Server Research Presentation​

In this meeting, guest speakers Lei Fu and Sahar Slimpour, from the Zurich University of Applied Sciences and University of Turku respectively, joined the CRWG to talk about their ROSBag MCP Server research (also shared in ROS Discourse).

The speakers discussed comparing different LLM models for analysing and reporting on recorded robot data. Due to how large robot datasets are, they often exceed LLM context limits, which makes the work challenging. To help, the speakers built a tool called MCP Lab which helped show different stats about LLM responses, such as time taken, tokens used, and so on.

The speakers also showed a live demo, where different LLM models could be selected to summarise robot performance or perform more complex tasks, like identifying if any objects came closer than a fixed distance over the dataset.

Finally, the CRWG asked questions about the research, such as how to trust answers from the LLM when it was occasionally wrong, and possible future features for the MCP Lab.

If you'd like to see the meeting, it is available on YouTube:

2025-09-08: Summer Holiday​

The group will not meet on this date, due to the number of members unable to attend. Meetings will resume from the 22nd September.

2025-08-25: Logging & Observability Review​

In this meeting, the CRWG discussed the similarities and differences between the tech of the companies that had come to talk about Logging & Observability. The group agreed that we need to talk to consumers of these technologies, so effort will be made to contact some of the companies with large fleets to request they talk to us. We can also release a summary of our findings so far to ask the community if they approve of the direction taken and the information gathered.

If you'd like to see the meeting, it is available on YouTube:

2025-08-11: Arun Venkatadri and Shouheng Yi present Bagel​

In this meeting, guest speakers Arun Venkatadri and Shouheng Yi presented a new open source project called Bagel. While both speakers are key members of Model-Prime, the Bagel project is not affiliated with Model-Prime.

Bagel lets you chat with your robotics data by using AI to search through recorded data. This is particularly relevant given the group has recently been researching Logging & Observability in preparation to write a community guide on the same subject.

Our guests spoke about the motivation behind the project before showing a technical demo, then answering questions from the group. In particular, we were interested in the pitfalls of using LLMs to understand structured data, and went into depth on this point. Arun has stated that the conversation inspired progress on the project and hopes to return in a future session to show a solution.

If you'd like to see the meeting, it is available on YouTube:

2025-07-28: Trying Out Heex and Discussing Anomaly Detection​

In this meeting, the group signed up for an account on the Heex Technologies website to test out the services provided. We read through the different features of the service, including creating a new robot, then worked through the accompanying documentation to generate fake high velocity events. The group then visualised the events with the service, learning more about how the triggers worked in the process.

If you'd like to see the meeting, it is available on YouTube:

2025-07-14: Guest Expert Bruno Mendes De Silva from Heex Technologies​

The group invited Bruno Mendes De Silva to join us at this meeting as a guest expert in Logging and Observability.

Bruno is the Co-Founder and CEO of Heex Technologies, which is a cloud data platform accelerating Physical AI adoption. The company helps customers with their autonomous systems by collecting and presenting only the most relevant data for a required view.

Bruno and Benoit Hozjan, who is a Project Manager in charge of customer experience, presented Heex Technologies together. First they discussed the company and the purpose of the service, then they demonstrated a showcase workspace for the visualisation and anomaly detection capabilities.

If you'd like to see the meeting, it is available on YouTube:

2025-06-30: Logging & Observability Discussion (Upcoming) CANCELLED​

This meeting is cancelled as the host is unwell. Meetings will resume as normal from the 14th July.

Previous text:

In the upcoming session, the group plans to go over the research performed to this date. This includes the first preparation session and the previous session, where we hosted a discussion with Benji Barash, Co-Founder and CEO of Roboto.AI.

This will be a quieter, more relaxed session, as some of the members are unable to attend.

2025-06-16: Guest Expert Benji Barash from Roboto AI​

The group invited guest expert Benji Barash, Co-Founder and CEO at Roboto AI, to discuss Logging & Observability.

Benji leads Roboto AI, which provides an analytics engine for physical AI. The company focuses on searching, transforming, and analyzing data from robots at scale. Previously, Benji led engineering and science teams at Amazon Robotics, with a focus on simulation and autonomy.

The meeting became more of a presentation and Q&A, as Benji had slides and demos ready to run. He showed very impressive tech from Roboto.AI, including the ability to select a clip of robot data and search other datasets for similar clips to find similar motions!

The full meeting recording is available below.

2025-06-02: Logging and Observability Preparation​

The group met to start researching Logging & Observability, with the intention of gathering information for a community guide on the subject. Collectively, they gathered the following points to consider going forwards:

  • Format for recording logs and data: structured data? Unstructured? Strings? JSON? Rosbags?
  • Online vs offline data upload: collecting data offline for uploading later
  • Automatic incident detection: is data collection constant, or triggered by a specific event, or possibly by a developer?
  • Message publishing frequency: a slower message publishing frequency often indicates a system issue.
  • Cause of Issue: determining the cause of the issue, if there is one.
  • Publishing Protocol: MQTT, ROS2 topics, HTTP, or other. Advantages and disadvantages of each method.
  • On demand vs constant logging
  • Interfacing with observability tools: publishing or formatting data to be compatible with observability tools such as Prometheus
  • Live visualisation tools: pros and cons of different tools such as RViz and Foxglove. For example, Foxglove can provide a web link, which is easier to share to other team members.
  • Levels of data upload: minimal upload such as basic logs; rosbags with debug data; developer-triggered upload of all robot information.
  • Custom downsampling: topics can be downsampled depending on the quantity of data and the bandwidth available. Another option is to extract important information, such as the closest lidar point or the classes recognised from the camera, to reduce the data uploaded.
  • Target user of data: who will make use of the data? This affects how the data is presented and used.

The group ended by discussing potential guest experts that could help provide information and insights for the guide.

If you would like to see the meeting recording, it is available on YouTube:

2025-05-19: Ideas for Cloud Robotics Guides​

The group met and used this time to discuss possible guides for publishing on Cloud Robotics Hub. This would make our expertise available to the public in an easily digestible format.

As part of the meeting, we agreed to post a poll to the community to get a better idea of priority on each suggested topic. The results are in this blog post.

If you would like to see the meeting recording, it is available on YouTube:

2025-05-05: Continued Discussion: Uploading Robot Data to the Cloud​

In this meeting, the group met to continue their discussion about uploading robot data to the cloud. They expanded from the previous discussion to include more important details about the data being uploaded, such as the moment it was captured versus an incident, to give one example.

The group also discussed difficulties with cloud data collection, such as the difficulty of automatic anomaly detection, and how companies frequently have to build their own data collection solution to meet their needs.

If you would like to see the meeting recording, it is available on YouTube:

2025-04-07: Discussion: Uploading Robot Data to the Cloud​

The group used this meeting to discuss uploading robot data to the cloud. Together, they worked through the most useful information to convey to robotics developers, resulting in the start of a Google Doc to organise the most important data by category of robotics.

The group finished by agreeing to continue the discussion in the following session.

If you would like to see the meeting recording, it is available on YouTube:

2025-03-24: General Catch-Up​

In this shorter session, two members met and discussed the use of LLMs in robotic and software development, then agreed on a new format that could be used in future sessions: proposing a discussion topic in advance and using the session to discuss that topic.

Based on this agreement, a new Github Issue can be used to propose discussion topics for future meetings. Anyone can post to the issue and join in the discussion.

If you would like to see the meeting recording, it is available on YouTube:

2025-02-24: Robot Deployment Discussion​

In this session, the group discussed their preferences and issues with the most popular deployment methods. The group agreed that distribution via Docker containers was the best method, but still had issues, such as the large download sizes of images. Later in the session, the group also discussed a possible project to remotely disable particular functions of a robot remotely.

If you would like to see the meeting recording, it is available on YouTube:

2025-02-10: Testing KubeEdge (For Real, This Time)​

The group had their second meeting to try out KubeEdge, having heard more about it from Tomoya Fujita in a guest talk.

Two of the members had some success with the prerequisites for KubeEdge, but neither were able to connect a Raspberry Pi to a cloud machine as an edge device. They were able to get a Kubernetes cluster connecting to another machine on the same network, treating it as an edge device and seeing ROS 2 traffic between the two machines.

The group summarised their experience of KubeEdge:

We found KubeEdge to be conceptually very powerful, and the functionality we did get working was very impressive. This looks to be an excellent way to manage robots or other edge devices at an enterprise level. However, we did have difficulties setting up KubeEdge for a Raspberry Pi at the edge. We think the following might be helpful to get others onboarded with KubeEdge:

  1. It would be great to have a complete reference setup. Set up an EC2 instance with an elastic IP, set up the K8s network from scratch, and get a Raspberry Pi (or VM on another machine) to join as an edge device.
  2. Some instructions in the setup documentation assume prior knowledge. For example, in the KubeEdge example led-raspberrypi, the last step is to change the device Twin attribute, but the instruction does not say how to do so.
  3. It would be good to have a way of setting up multicast for ROS 2 that doesn't require changes at the cluster admin level to adjust for changes in the ROS 2 network. For example, if the domain ID changes, a new multicast address must be enabled for the cluster, which requires cluster admin rights. The alternative is to configure both the multicast IP address and subscriber IP addresses for each pod, which is more complicated.

If you would like to see the meeting recording, it is available on YouTube:

2025-01-27: Finding KubeEdge Prerequisites​

This meeting, the group planned to meet to test out KubeEdge. After the meeting started, the group realised that there were a number of prerequisites needed to try KubeEdge, including an active Kubernetes cluster (preferably in the cloud) and some hardware to deploy to, such as a Raspberry Pi with LED.

The group agreed to set up the prerequisites and attempt the tryout session in the next session. The group also agreed that for future tryout sessions, they would read through prerequisites in advance to make sure the meeting can go ahead.

If you would like to watch the meeting, you can see the recording below.

2025-01-13​

The group had a general discussion of the latest in robotics news, including NVIDIA Cosmos, the proliferation of humanoid robotics at present, what cloud robotics encompasses, and what to do in the next meeting. The group settled on doing a KubeEdge trial unless a guest speaker is available.

If you would like to watch the meeting, you can see the recording below.

2024-12-16 Guest Talk: Robotics Platform empowered by Cloud-Native Technologies (Tomoya Fujita)​

In this meeting, we hosted a talk by by Tomoya Fujita, a software engineer, contributor to ROS2 core, and well-respected community builder.

Tomoya talked about a range of Cloud-Native Technologies he has worked on, and their relevance to robotics. In particular, Tomoya has done a great deal of work on KubeEdge, a platform built on Kubernetes that can be used to deploy software to robots as if they are Kubernetes clusters. Some resources from his talk include:

More information on the talk is available on the blog post, including a link to the recording posted on YouTube.

If you are interested in seeing the entire meeting, the recording is available here:

2024-12-02 | Testing Eclipse Zenoh​

In this meeting, the group spent the time trying out the workshop from Eclipse Zenoh that was used in the latest ROSCon. Together, they accomplished a few of the exercises and discussed the ease of setup, and some comments they could give as feedback.

The workshop is available from Github. The group completed exercises 1, 2, and 5, then two members sent messages to each other using the instructions in exercise 3.

The meeting recording is available below:

2024-11-18​

The group caught up on the latest robotics news and discussed GenAI in robotics and the start-ups using them. The group then agreed to focus on Eclipse Zenoh for its next meeting.

If you would like to watch the meeting, you can see the recording below.

2024-11-04 Guest Talk: Eclipse Zenoh and its offering for Cloud Connectivity (Julien Enoch)​

The group hosted a talk by Julien Enoch, a Senior Solutions Architect at ZettaScale Technology and Eclipse Zenoh committer.

Julien spoke about the Zenoh protocol, which can run everywhere, from micro-controllers to the Cloud; over TCP, UDP, QUIC, and Websockets. Zenoh provides a software router that can be deployed in any Cloud instance such as AWS EC2, and that can route the protocol between different subsystems. The talk also covered how this is particularly helpful for Connectivity in Cloud Robotics, with a video demonstration of controlling a robot in California from an RViz instance in France!

If you would like to see the talk, which took the whole meeting time, see the recording below.

2024-10-21​

The group discussed recent conferences, possible resources for the Cloud Robotics Hub, and methods of working offline to accomplish more between meetings. The group also discussed their opinions on how the Working Group was doing, with the aim of recognising their successes and considering how to improve going forwards.

The full meeting recording is available below.

2024-10-07 Guest Talk: Enabling Usable and Reliable Cloud Robotics with FogROS2 (Kaiyuan Eric Chen)​

The group hosted a talk about FogROS2 presented by Kaiyuan Eric Chen, a PhD student in the Department of Computer Sciences at UC Berkeley and a member of Berkeley Automation Lab working closely with Ken Goldberg and Jeffrey Ichnowski.

More information about the presentation, including a summary of its content and some links, are available on the blog post.

The group spent some time at the end discussing recent news in robotics, and the plan for the next meeting.

The full meeting recording is available below.

2024-09-23 Talk: CloudGripper (Florian Pokorny)​

The group hosted a talk about CloudGripper, An Open Source Cloud Robotics Testbed for Robotic Manipulation Research, Benchmarking and Data Collection at Scale by Florian Pokorny, Associate Professor of Machine Learning at KTH Royal Institute Technology and Co-Founder of Scaleup Robotics.

Here's a link to the blog post and talk recording.

Find the full meeting recording below.

2024-09-09​

The group discussed the launch of the new Cloud Robotics Hub site and discussed potential improvements. Akash Vibhute from Intrinsic/OpenRMF attended and we discussed him possibly presenting to the group, potentially along the lines of his upcoming ROSCon talk.

2024-08-26​

This meeting, the group discussed updates on the hub site, then talked about the work occurring in academia for cloud robotics.

2024-08-12​

This meeting, the group took a look at progress on the Cloud Robotics Hub website, then agreed tasks to split between themselves to get the site closer to publishing.

2024-07-29 Talk: ROS Tool (Christian Fritz)​

This meeting, the group got a first look at the upcoming Cloud Robotics Hub website, then heard about rostool from group member Christian Fritz of Transitive Robotics.

rostool Demo and Explanation -- Cloud Robotics Working Group 2024-07-29

from Open Robotics on Vimeo.

2024-07-15​

This meeting was the first meeting following the new format. The group members shared news about the upcoming Cloud Robotics Hub site and discussed methods of measuring latency between sending commands and the robot responding.

No meeting recording is available for this meeting.

2024-07-01​

In this meeting, we had a discussion about the purpose of the working group going forwards. We decided to change the meeting format to focus less on group member actions and more on being a space for cloud robotics enthusiasts can share news with each other, with more guest speakers.

No meeting recording is available for this meeting.

2024-06-17​

We hosted a guest in this meeting, with Gui Manzato from Ekumen talking about his experience working with all layers of the robotics stack and contributing back to ROS and open source software.

No meeting recording is available for this meeting.

2024-06-03​

This meeting was used to continue processing results from the State of Cloud Robotics Survey.

Cloud Robotics Community Group Meeting 2024-06-03

from Open Robotics on Vimeo.

2024-05-20​

This meeting, we continued to process our survey results, and created an initial set of results to share with the community. The resulting LinkedIn post is available here.

No meeting recording is available for this meeting.

2024-05-06​

This meeting was the first meeting we started to use to process our survey results.

Cloud Robotics Community Group 2024-05-06

from Open Robotics on Vimeo.

2024-04-22​

We used this meeting to continue drafting the questionnaire from the previous meeting.

Cloud Robotics WG Meeting 2024-04-22 | Building State of Cloud Robotics Survey

from Open Robotics on Vimeo.

2024-04-08​

We used this meeting to discuss how to gather data from folks working with cloud robotics. We agreed to gather data using a public questionnaire and began drafting it. The draft is available here.

No meeting recording is available for this meeting.

2024-03-11​

In this meeting, we discussed our long-term strategy and built a set of slides that laid out that strategy. The slides are available here.

Cloud Robotics Working Group Meeting March 11th, 2024

from Open Robotics on Vimeo.

2024-02-26 Talk: Asimovo (Christine Fraser)​

This was our first meeting as a community working group. We introduced ourselves to each other, then had a presentation from Christine Fraser and Ludo Stellingwerff from Asimovo.

ROS Cloud Robotics Working Group Meeting February 25th, 2025

from Open Robotics on Vimeo.