jaxls

Project notes

Try the waypoint problem before a robot model

The included examples/3_constraints.py builds 50 two-dimensional waypoints, fixes the start and goal, and adds constraints around two circular obstacles. It solves from a straight-line initial guess, prints constraint violations and displays the initial and optimized paths in Viser. This makes a small, inspectable starting point for understanding how costs and constraints are assembled. [3]

For a source-based trial, use an isolated Python 3.10+ environment, check out the reviewed revision and install with pip install -e ".[dev]"; the dev extra supplies Viser. Run python examples/3_constraints.py from the checkout. These steps are derived from the package metadata and example and have not been run for this review. JAX and jaxlib must be at least 0.6.0. A workload-specific memory or accelerator minimum is not established here. [2] [3]

The example checks obstacle constraints at discrete waypoints. It does not model a robot's volume or test the swept segments between those points. Treat its printed success message as a check of that mathematical example, not evidence of safe physical motion. [3]

Record the commit, not just the package version

No GitHub release was listed at review. This profile uses commit 50a58be88c5ef74532f09e3f55268b4f02c490e3; the package still identifies itself as 0.0.0. Its compatibility shims warn that Factor became Cost and FactorGraph.make(...) became LeastSquaresProblem(...).analyze(). Record the resolved commit and JAX environment when comparing results or integrating an older consumer. A README relationship alone does not verify version compatibility. [4] [6]

Repository code and documentation carry MIT terms; dependencies keep their own licenses. We preserved and inspected source but did not install jaxls, run its examples or benchmark a solver. [7]