Skip to content

Online GCS

Online GCS is a Python 3.12 research library for manipulation motion planning. Its standard run evaluates an RRT baseline on every query and a TrajOpt baseline when that path exists, then checks a Graph of Convex Sets (GCS). If the current cover is insufficient, an RRT path supplies keypoints for new IRIS regions and a GCS retry. Baseline runtime is therefore present even on GCS successes, and TrajOpt depends on a compatible solver.

Online GCS planning pipeline

Start here

  • Installation — create a supported environment and install extras.
  • Quick start — run the CLI, Python API, examples, and three scenes.
  • Architecture — understand GCS, RRT*, keypoints, IRIS, and graph updates.
  • Python API — supported public classes and generated reference.
  • CLI reference — commands, modes, and validation rules.
  • Reproducibility — fixed seeds, checks, and output policy.
  • Troubleshooting — models, solvers, Meshcat, and runtime issues.

Release boundary

Version 0.1.0 is a pre-1.0 research release. It supports Python 3.12 and keeps reports, notebooks, raw experiment results, logs, and generated data outside the public tree. See the changelog for release details.