Managing configs is a pesky infrastructure problem that slows down scaling robotics fleets.

A single robot can have thousands of configuration parameters, many of which are custom, based on factors such as the customer site, hardware version, etc. This makes it difficult to manage which configs belong on which robot, version them globally (while maintaining sanity for local edits), and deploy them to your device.

In this blog, I wrote about the challenges teams have faced with config management and how we’ve designed Miru to solve them.