1 of 9

Robotics Cloud WG Long-Term Strategy

2024-03-11

2 of 9

Cloud Robotics

  • Valid use cases (non-exhaustive list):
    • Augment self-sufficient robots
    • Run robot applications directly in the cloud
    • Cloud tools to support robotic applications, e.g. simulation
    • Support edge compute in style of cloud, or change behavior depending on network awareness e.g. reduced points in point cloud

3 of 9

Tenets

  • Not just cloud roboticists: our end users are pure cloud developers, pure roboticists, AND cloud roboticists. Our content is accessible to everyone.�
  • Open by default: we publicly release any data we gather or tools/resources we build, for free.�
  • Transparent: we share everything using public-facing website. For example, text discussions will be via ROS Discourse, and any video meetings will be recorded and uploaded to a public channel.

  • [POSSIBLE] Safety and Security: we consider safety and security of the highest priority while writing our tools.

4 of 9

Phases

  • Data Collection
  • Data Analysis
  • Build & Scale

5 of 9

Data Collection Phase

  • Gather data on questions such as:
    • What's easy and hard about cloud robotics?
    • How are people solving the challenges right now? Who are they?
    • What are the pros and cons of connecting robots to the cloud?
    • What technology sections are there for us to focus on?
    • Who is already building technology for this, and what are they doing?
    • What resources are available online for making cloud robotics better?

  • Gather data by:
    • Knowledge from our own members
    • Performing research
    • Inviting guests to speak

  • Track and identify tools that we can use in the Build & Scale Phase

6 of 9

Identified Technology Categories

  • While discussing the data collection phase, we identified the following technology categories:
    • Data collection & streaming
      • Point clouds in particular have proven very difficult
    • Teleoperation
    • Remote monitoring and/or control
    • Artificial Intelligence
    • Communication patterns
      • Pub/sub vs service call
      • Symmetric vs asymmetric
    • Connection speed, latency, jitter, and bandwidth
    • Offload algorithms to the cloud or edge

7 of 9

Data Analysis Phase

  • Categorize for focus areas: builders, consumers, and technology categories
  • Build "are we _ yet" style website for Cloud Robotics (see next slide)
    • Open to everyone
    • Accept contributions
    • Useful for cloud-only and robot-only alike
  • Identify reason for current state, e.g. why is MQTT the de facto standard? Is RoboMaker the best tool for simulation?
  • Identify experts in different domains
  • Identify resources for users, such as available tooling and guidelines for users in particular area, and cost of options
  • Identify problem areas for developing Cloud Robotics
  • Build a plan/roadmap for tackling these areas

8 of 9

Are we _ yet?

9 of 9

Build & Scale Up Phase

  • Build
    • Follow the roadmap to develop problem areas
    • Keep "are we _ yet" page up to date
    • Encourage non-members to use our data e.g. build tools, start companies
    • Offer to own & maintain open source, or create our own, to bridge gaps
    • Consider becoming full TSC WG as we build and grow
    • Consider how we can influence ROS2 itself to make it more cloud compatible�
  • Scale Up
    • Reach out to larger robotics companies; pure cloud developers; pure roboticists
    • Show what we've built (info page, existing tools)
    • Encourage them to help and invite them to the WG
    • Build a network of contacts
    • Reach out to other developers to build a Cloud Robotics community