v0.4.0 - FCIT* Release

This release includes the implementation of the Fully-Connected Informed Tree (FCIT*) planner from https://arxiv.org/pdf/2411.17902.
This is an asymptotically optimal motion planner that provides state-of-the-art performance on the MotionBenchMaker dataset.
There have also been changes to the RNG API to enable better randomization of multiple trials.

What's Changed

• Fix typo bug with vreinterpretq + CI improvements by @wbthomason in #21
• Add support for filtering NDArray pointclouds by @wbthomason in #23
• fix: explicitly use float32 for numpy typing by @zkingston in #27
• feat: RNG is now a planner argument, expose XORShift generation by @zkingston in #29
• Add the FCIT* asymptotically optimal motion planner by @zkingston in #35

Full Changelog: v0.3.0...v0.4.0

v0.3.0 - End-effector Attachments

What's Changed

• Add support to add custom end-effector attachments for Panda, UR5, and Fetch in #18
• Example OMPL Integration Script in #16
• Various bug fixes

Attachments

See the scripts/atttachments.py for an example of how to attach custom geometry to the end-effectors of the UR5, Panda, and Fetch robots.

Initial CAPT Release

This release contains the implementation of the Collision-Affording Point Tree (CAPT) from the forthcoming RSS 2024 paper “Collision-Affording Point Trees: SIMD-Amenable Nearest Neighbors for Fast Collision Checking” (http://arxiv.org/abs/2406.02807). The CAPT enables fast collision checking of our spherized robot models against pointcloud data, and has an average query time of less than 10 nanoseconds on 3D scenes composed of thousands of points.

Initial Release

Initial public release of VAMP

Co-authored-by: Wil Thomason <[email protected]>