Click here for the documentation: https://docs.trossenrobotics.com/trossen_arm/main/tutorials/openpi.html We’re excited to announce that Trossen AI hardware is now fully integrated into the OpenPI framework, making it easier than ever to collect data, train Pi0 and Pi 0.5 policies, and inference— all in one unified ecosystem. OpenPi, developed by the team at Physical Intelligence, is an open-source vision-language-action model. With our new dedicated fork, you can now: • Collect episodes using your Trossen AI hardware • Train and fine-tune Pi0 and Pi 0.5 policies on your own datasets • Run inference right out of the box • Follow detailed documentation to streamline setup, training, and evaluation This integration is just the beginning. A complete walkthrough is coming soon, where we’ll show you how to go from collected data sets to a fully deployed model using your Trossen AI hardware. Make sure to subscribe so you don’t miss it. https://www.trossenrobotics.com

To learn more about the WidowX AI, or if you are ready to purchase, please visit www.trossenrobotics.com/widowx-ai

To learn more about the WidowX AI, or if you are ready to purchase, please visit www.trossenrobotics.com/widowx-ai

To learn more about the WidowX AI, or if you are ready to purchase, please visit www.trossenrobotics.com/widowx-ai

Learn how to tweak driver parameters to your project's needs. https://docs.trossenrobotics.com/trossen_arm/main/getting_started/configuration.html

Join Shantanu as he walks you through the seamless integration of the Trossen AI lineup with the Huggin Face LeRobot data pipeline. In this tutorial, you’ll see how to leverage the full potential of your Trossen hardware using LeRobot’s robust ecosystem of machine learning models and data processing tools. LeRobot offers a comprehensive suite for data ingestion, transformation, and annotation, making it the ideal platform for building and refining ML models with real-world robotics data. Key features include: ✅ Access to a diverse range of ML models – from supervised to reinforcement learning ✅ Streamlined data pipelines for managing large-scale robotic datasets ✅ Built-in tools for automated data labeling, model training, and evaluation ✅ Interoperability with major ML frameworks like PyTorch and TensorFlow ✅ Extensive API and SDK support for customizing your workflows Whether you’re developing advanced perception systems, refining motion control, or exploring reinforcement learning, LeRobot empowers you to harness your data effectively and accelerate your robotics research and product development. Watch now to learn how to get started and unlock the full capabilities of your Trossen AI lineup with LeRobot!

Shantanu covers the first of many simulation environments we’ll be rolling out this summer. First up is MuJoCo, the physics engine known for its exceptionally fast and accurate simulation of contact-rich robotic interactions. What sets MuJoCo apart from NVIDIA Isaac, Gazebo, Grid, and other environments is its deep focus on high-fidelity physics for real-time control and model-based reinforcement learning. MuJoCo’s continuous-time formulation and advanced optimization tools make it particularly powerful for simulating soft contacts and complex multi-body dynamics, which are crucial for training and validating advanced AI models. We’re excited to bring this powerful tool to the Trossen AI lineup, enabling researchers and developers to test and optimize control policies with precision. Stay tuned for more simulation rollouts this summer! #robotics #simulation #MuJoCo #machinelearning #TrossenAI

Shantanu goes over the Trossen AI firmware version 1.8, released in late spring 2025. For the latest documentation and firmware visit https://docs.trossenrobotics.com/trossen_arm/main/index.html