Skip to content

Latest commit

 

History

History
65 lines (49 loc) · 3.46 KB

File metadata and controls

65 lines (49 loc) · 3.46 KB

Project dependencies

O3DE

Repository URL
https://github.com/RobotecAI/o3de.git

O3DE gems

Gem Name Repository URL
o3de-extras https://github.com/robotecai/o3de-extras.git
o3de-humanworker-gem https://github.com/RobotecAI/o3de-humanworker-gem.git
o3de-ur-robots-gem https://github.com/RobotecAI/o3de-ur-robots-gem.git
robotec-o3de-tools https://github.com/RobotecAI/robotec-o3de-tools.git
robotec-warehouse-assets https://github.com/RobotecAI/robotec-warehouse-assets.git
robotec-generic-assets https://github.com/RobotecAI/robotec-generic-assets.git

ROS 2

Most ROS 2 dependencies (MoveIt 2, Nav2, ros2_control, control_msgs, simulation_interfaces, and the rest) come from the RoboStack robostack-jazzy conda channel and are installed automatically by pixi — see [dependencies] in pixi.toml. They are no longer vcs-imported or built from source.

These packages are built from source (see ros2_ws.repos):

Package Name Repository URL
robotnik_common https://github.com/RobotnikAutomation/robotnik_common.git
robotnik_description https://github.com/RobotnikAutomation/robotnik_description.git
robotnik_sensors https://github.com/RobotnikAutomation/robotnik_sensors.git
rai_interfaces https://github.com/RobotecAI/rai_interfaces.git

Python

Package Name Repository URL
rai-core https://github.com/RobotecAI/rai.git
rai-whoami https://github.com/RobotecAI/rai/tree/main/src/rai_whoami

Inference

Inference Engine Backend Repository URL
llama.cpp GPU (Vulkan)/CPU https://github.com/ggml-org/llama.cpp.git
FastFlowLM AMD Ryzen AI NPU https://github.com/RobotecAI/FastFlowLM.git

Models

config.toml is the single source of truth for which models are served, on which backend, and from which weights. The endpoints it defines today:

Model name Type Backend Source
gpt-oss-20b LLM GPU (llama.cpp) https://huggingface.co/openai/gpt-oss-20b
LFM2-VL-3B VLM GPU (llama.cpp) https://huggingface.co/LiquidAI/LFM2-VL-3B-GGUF
gemma3:4b VLM NPU (FastFlowLM) https://huggingface.co/google/gemma-3-4b-it
Qwen3-Embedding-0.6B Embedding GPU (llama.cpp) https://huggingface.co/Qwen/Qwen3-Embedding-0.6B
Qwen3-Reranker-0.6B Reranker GPU (llama.cpp) https://huggingface.co/Qwen/Qwen3-Reranker-0.6B

LFM2-VL-3B serves the inspection, general, megamind, and condition agents; gemma3:4b serves the safety agent on the NPU.