This repository consolidates legacy core-*, shared_messages, and ts-spatial sub-projects into one import namespace: dexim.core.

This repository uses a root Pixi workspace with Python 3.10 as the baseline. pinocchio is installed from conda-forge in the same environment. For local development, the root pixi.toml solves only the host platform (currently win-64).

1. Install pixi (if not already installed):

# Windows (PowerShell) iwr -useb https://pixi.sh/install.ps1 | iex

# Linux / macOS curl -fsSL https://pixi.sh/install.sh | sh

See https://pixi.sh/latest/#installation for other options.

2. Resolve the root environment:

pixi install

3. Validate imports:

pixi run check-import

4. Optional: install PyTorch (CUDA 13.0). This is not managed by this project:

pixi run pip install torch torchvision --index-url https://download.pytorch.org/whl/cu130

For other CUDA versions or CPU-only builds, see https://pytorch.org/get-started/locally/

• dexim.core.config
• dexim.core.model
• dexim.core.nodes
• dexim.core.robot_interface
• dexim.core.utility
• dexim.core.messages
• dexim.core.spatial

The legacy folders under src/dexim/core were standalone packages (each with its own pyproject.toml). The extracted layout keeps those capabilities but exposes them as modules in a single package so you can install and import from one place.

from dexim.core.config import ControlNodeConfig from dexim.core.nodes import ManagedNode from dexim.core.messages import TopicBuilder from dexim.core.spatial import Transform3D