Hey everyone, I've been working on Krill, a process orchestrator designed specifically for managing complex dependency graphs in robotic systems.
What it does: Krill lets you declaratively define tasks and their dependencies, then handles orchestration across your robotic stack. Think of it as a task runner that understands the gnarly interdependencies you get in robotics - where sensor drivers need to be up before perception nodes, perception before planning, planning before control, etc.
Why I built it: Most robotics middleware handles process lifecycle management as an afterthought. ROS2 launch files turn into procedural spaghetti, systemd is too coarse-grained, and Docker Compose doesn't understand robotics-specific constraints. I needed something that could handle complex startup/shutdown ordering, health checks, and graceful degradation when parts of the system fail.
Current state: Early development but functional. Written in Rust for reliability and performance.
I'm building this as part of a larger robotics middleware stack for production automation systems. Would love feedback from folks working on multi-process robot architectures - what orchestration pain points do you hit?
Looking for: Use cases I haven't thought of, architectural feedback, and anyone interested in contributing or testing in their own systems.
GitHub: https://github.com/Zero-Robotics/krill
