3 posts tagged with "cloud"

This post shows how to deploy an EC2 Spot Instance on AWS capable of running O3DE, Ignition Gazebo, and NVIDIA Isaac Sim at a reduced cost. This Spot Instance is part of a system that uses an Amazon S3 bucket to store project data, such as robot environment levels, and scripts to automate starting up and shutting down instances to save on costs when the simulation instance is not in use; the whole system is deployed using CDK, except for a few manual setup steps.

By the end of this post, you’ll have a fully automated system to spin up high-end GPU simulation machines for a fraction of the usual cost. What's more, you'll have ROS 2 access, remote desktop viewing, and one-line start/stop scripts to manage the instances.

This post is also available in video form. If you'd prefer to watch, click the link below:

AWS IoT Greengrass, or Greengrass for short, is a way of deploying software to an edge device, like a server on a factory floor, and managing the lifecycle of that software. From the cloud you can configure the software that is deployed to the edge and how that software is configured, as well as build your own software to be deployed by Greengrass.

There are three parts: building and deploying software applications, the device(s) that those applications are being deployed to, and the management service doing the deployment. This post goes over each part to explain how Greengrass as a whole works, and how you can try it out for yourself.

This post was originally published as a video, but given how many people it has helped, converting it to a blog post made sense. If you prefer video format, click below.

This is a crucial question when deciding whether to use the cloud at all for robotics development, let alone how much to use it. Why connect my robots to the cloud at all? What benefits does it bring me, and what trade-offs am I making in order to use the cloud? This post is all about the why, rather than the how.

If you want the short answer, here it is: you should use the cloud if you ever plan to scale your robot fleet past around 10 robots, and if you do intend to scale up, integrate with the cloud as early as possible to avoid integration pains. If you're never intending to scale that much, the cloud can still be of use, but it depends on your use case as to whether the benefits outweigh the costs.

If you're not convinced, read on! The rest of this post is explaining my statement above.