Software
Robotic automation in research encounters challenges like fragmented hardware ecosystems, non-reusable solutions tied to specific robots, and knowledge loss due to researcher turnover. Efforts on non-use-case-specific tasks, often requiring automation engineers, further hinder research. Existing ecosystems lack modern technology utilization, impeding complex programming and collaboration.
The Platform
To address these issues, we've created software abstractions and corresponding adapters for our lab's hardware. Each abstraction is paired with a server running on our on-premise cluster and a lightweight client for remote access. Clients can run on various platforms, offer soft-real-time capabilities, especially control loops involving a multitude of sensors, actors and robots, at 100-200Hz. Solutions are hardware-agnostic, encapsulated with user-friendly high-level APIs, making them accessible to non-experts.
IRZ Robot Playground
Our lab houses an extensive array of interconnected robots, sensors, and systems, all controlled through the servers on a VMware cluster. With the Python client, installable via pip, students can manipulate robots and access sensor data within hours, promoting seamless integration with other modules in the ecosystem. Below, the creation of an automated experiment in WAAM making use of an early version of the platform, and the result of a 3 day hack, where a hand-full of IRZ people celebrated the last days before Christmas 2022.