MCP Cross-Robot Goal Creation

Creating robots that work together and collaborate to accomplish a bigger goal? Cadenza Lab NOW OFFERS YOU THE ABILITY TO DO THAT!!!

Use our MCP server, connect each robot to the MCP, set up a gym with both robots, and then RUN! Enjoy hacking!

1.5.1: Bug Fixes in LoRA

Improved bugs allows for easier fine-tuning and manipulation of the LoRA to different circumstances with both weight- and text-based fine-tuning, fulfilling the full-stack improvement abilities!

1.5.0: LoRA action control for VLA

Improving the VLA model's ability to interface with action library using a LoRA layer. INCREASE CUSTOMIZABILITY for you to do RL/fine-tuning Cadenza-based tech-stacks.

Happy building!

1.4.1: Bug fixes + New Tests

Enjoy a STREAMLINED developer experience with bug fixes AND new tests to try out!

3D memory has also been ADDED for y'all to enjoy! Make your AI models enjoy the ability to understand their 3D environments more into depth, with new features being launched regularly!

1.4.0 - Model Orchestration + DX improvements

Build much more smoothly with the new ORCHESTRATION LAYER! Setup how you want your on-board AI to work and make your robots smarter, whether you want it to be a preset action with small pivots on the way... or FULL ON-EDGE movement!

TRY NOW! https://pypi.org/project/cadenza-lab/1.4.0/

1.3.3/4 - Custom Gyms & Bug Fix

Need to build your own gyms for testing your physical AI tech stacks more effectively? Cadenza Lab officially supports that, allowing you to test your system to YOUR circumstances and situation.

Check it out!
https://pypi.org/project/cadenza-lab/1.3.4/

1.3.1

Improved pip package setup. Just run 'pip install cadenza-lab' to call and use your favorite Cadenza features!

BIG CHANGE: Added VLA Model Integration into Cadenza, reducing the "blackbox-ness" of training physical AI models.

Just use the following and enjoy!

policy = SmolVLAPolicy.from_pretrained("lerobot/smolvla_base")

cadenza.stack.register_world_model("smolvla", model=policy)
cadenza.stack.run(robot="go1", goal=("walk_forward then climb_step, then walk_forward then sit"), target=(-5.5, 0.0), xml_path=SCENE, max_iterations=40)

1.3.2

NOW COMBINE YOUR OWN MODELS to build effective physical AI stacks ON CADENZA!

Multi-modal sensing added to help your inserted VLA model process data (e.g. depth estimation) and make effective actions.

Along so, a NEW ACTION LOOP is integrated, allowing your VLA model to communicate in Cadenza's action library and is proactive, allowing for easier debugging and improve RL systems.

import cadenza_lab as cadenza
import cadenza_lab.stack
from depth_anything_v2_small import DepthAnythingV2Small
from lerobot.policies.smolvla.modeling_smolvla import SmolVLAPolicy

cadenza.stack.register_world_model(
"smolvla", model=SmolVLAPolicy.from_pretrained("lerobot/smolvla_base"),
)

cadenza.stack.run(
robot="go1",
goal="reach the green beacon at the top of the stairs and sit",
target=(-5.5, 0.0),
xml_path="examples/smolvla-test/stairs_scene.xml",
modalities=[DepthAnythingV2Small()],
)

Full Changelog: https://github.com/aparekh02/cadenza/commits/latest

1.2.1

Announcing the FIRST simple Robotics SDK to run and deploy complex physical AI operations. Install this lightweight package and BUILD YOUR PHYSICAL AI FLOWS!!

1.2.2/3

Cadenza has released its FIRST PIP PACKAGE: cadenza-lab!

The command 'pip install cadenza-lab==1.2.3' makes this system much easier to use in larger deployment environments and scaling our updates more effectively with pip's update feature.

Happy coding with Cadenza!