Changelog

Contents

Changelog#

v0.1.0 (Current)#

Initial release.

  • Three-layer modular architecture (Abstract Interfaces → Concrete Implementations → Task Integration)

  • DRIS as universal state representation with domain randomization support

  • Cage-constrained action selection via ManiDreamsEnv.dream()

  • Simulation-based and learning-based TSIP backends

  • Solvers: PolicySampler (RL), MPPIOptimizer (planning), GeometricOptimizer (discrete)

  • Cage implementations: CircularCage, DRISCage, PlateCage, Geometric3DCage, CircularPixelCage

  • Five task categories: object pushing, catching, picking, pixel-based pushing, ManiSkill defaults

  • Documentation site with Sphinx + sphinx-book-theme, deployed via GitHub Pages

Contents