Built for Sim-to-Real Robotics Learning

The Test Tube Manipulation Education Kit provides the physical props used in NVIDIA Sim-to-Real robotics learning workflows, including centrifuge vials and a yellow tube rack.

Designed for use with reBot and SO-ARM101, the kit creates a standardized tabletop manipulation scenario where learners can practice robotic grasping, pick-and-place, and test tube manipulation while exploring NVIDIA Isaac Sim and Sim-to-Real workflows. Follow the reBot Sim-to-Real Course or the SO-ARM101 Sim-to-Real Course to get started.

What Is Sim-to-Real?

Sim-to-Real is the process of developing or training robot policies in simulation and transferring them to physical robots.

With a consistent manipulation setup in both simulation and the real world, learners can explore how differences in perception, physics, robot dynamics, and the environment affect real-world robot performance.

Learn with NVIDIA Isaac Sim

NVIDIA Isaac Sim is a robotics simulation platform built on NVIDIA Omniverse. It provides a physics-based virtual environment for developing, testing, and training AI-powered robots before deploying them on physical hardware.

By recreating the test tube manipulation task in Isaac Sim and using the corresponding physical props in the real world, learners can build a practical Simulation → Training → Real-World Deployment workflow.

Sim-to-Real with SO-ARM101

Pair the kit with SO-ARM101 and follow NVIDIA's Sim-to-Real learning course to explore the complete workflow from simulation to physical robot deployment.

The standardized test tube task provides a practical environment for learning:

• Isaac Sim Environment Setup — Recreates the robot and manipulation task in simulation.
• Robot Manipulation — Practice grasping, moving, and placing centrifuge vials.
• Sim-to-Real Transfer — Transfer workflows developed in simulation to the physical SO-ARM101.
• Policy Evaluation — Compare robot behavior between simulated and real environments.

Free Physical AI & VLA Course

In collaboration with NVIDIA, Seeed offers a free Physical AI & VLA course using the Test Tube Manipulation Education Kit as a standardized training setup. Learn to collect real and simulated robot data, augment datasets with NVIDIA Cosmos, fine-tune Isaac GR00T 1.7 VLA models, and deploy them with TensorRT on Jetson.

Course Highlights:

• Real & Sim Data Collection — Collect teleoperation data with the reBot B601-RS and simulate SO-ARM101 setups in NVIDIA Isaac Sim.
• Cosmos Scene Augmentation — Generate diverse training data with NVIDIA Cosmos.
• GR00T 1.7 Fine-Tuning — Fine-tune VLA models on custom robot datasets.
• Jetson Deployment — Accelerate VLA inference with TensorRT on Jetson platform.

Array