ManfredCh/Lyapunov
LLM-first, 3D-native robotics and simulation workspace.
What it solves
Lyapunov provides a unified desktop workspace for robotics developers and 3D researchers to manage the entire lifecycle of a simulation project. It eliminates the friction of switching between separate tools for scene construction, robot preparation, physics simulation, and AI-driven agent control.
How it works
Built on an "everything-is-a-plugin" architecture, the framework integrates a 3D viewer, code editor, and terminal into a single interface. It supports multiple physics engines (such as MuJoCo and Isaac Sim) and allows users to either use a cloud-based agent (Peiri) for orchestration or configure their own local model providers in a guest mode. Users prepare robots and environments by importing standard formats (like GLB, MJCF, or URDF), configuring physics properties, and executing bounded actions within a controlled simulation loop.
Who it’s for
It is designed for robotics engineers, simulation research teams, and 3D creators who need a centralized environment to build, test, and observe robotic behaviors and 3D scenes.
Highlights
- Unified Workspace: Combines chat, file management, 3D visualization, and physics control in one desktop application.
- Multi-Engine Support: Integrates with MuJoCo, Isaac Sim, and Newton for diverse simulation requirements.
- Agent-First Workflow: Uses LLMs to orchestrate scene generation, robot preparation, and simulation tasks via natural language.
- Portable Projects: Supports saving and reopening projects with complete resource closure, ensuring consistency across sessions.
Related
- Project
- Project
- Project
- Project