Create a robot pick-and-place simulation using SimReady assets.
Isaac Sim 6.2 and SimReady assets enable a reusable robot arm and gripper pick-and-place test, checking joints, motion, and contact before use.
Key Takeaways
- The walkthrough builds a beverage-can pick-and-place simulation in Isaac Sim 6.2 using SimReady OpenUSD assets for a UR17G robot arm, Robotiq 2F-140 gripper, and a can.
- SimReady assets provide validated simulation-ready properties, such as joints, drive settings, collision, and mass, enabling more reliable testing before running manipulation tasks.
- The workflow has three main steps: download and inspect assets, assemble the robot and scene, then test controls and run the pick-and-place sequence.
- GPT-6 Sol is used as an agent-assisted coding path, but the core value is the reusable SimReady asset setup rather than proving real-world robot performance.
Meta description: A walkthrough shows how Isaac Sim 6.2, SimReady Central assets, UR17G, Robotiq 2F-140 and GPT-6 Sol build a beverage-can pick-and-place scene.
Tags: SimReady Central, Isaac Sim 6.2, pick-and-place, GPT-6 Sol, UR17G
Featured image: Abstract, friendly illustration style showing a simplified robotic arm, gripper, and beverage can in a clean simulation scene; no real people, no logos, no text.
Robotics developers and simulation builders can use a SimReady pick-and-place workflow in Isaac Sim 6.2 to move a beverage can with a UR17G arm and a Robotiq 2F-140 gripper. The example also uses GPT-6 Sol as an agent-assisted coding path, but the main value is the reusable asset setup: download validated OpenUSD packages, assemble the robot, and test joints, movement, and contact before running a manipulation task. The result is not a real-world robot claim; it is a repeatable way to check whether simulation-ready assets behave as expected inside a scripted scene.
What SimReady Adds to the Robot Scene
SimReady is a label for an OpenUSD asset that has been prepared for a defined simulation use case. In this workflow, that means the UR17G arm and Robotiq 2F-140 gripper include joints and drive settings, while the beverage can includes collision and mass properties. Those properties decide whether a robot that looks correct while stopped can also move, grasp, and stay in contact with objects when physics runs.
SimReady Central, hosted on Hugging Face, is the open library for these assets. It provides robot and object models with authored simulation properties and profile-validation evidence. SimReady Foundation defines the rules behind that evidence through three connected levels: profile, features, and requirements. A profile groups the features needed for a use case, such as a robot body. Features describe capabilities, such as controllable joints. Requirements check the settings needed for each feature, such as a joint's drive settings.
For a user, the practical meaning is reuse with a starting point. The asset packages contain the models and their simulation settings. The user still assembles the scene and adds task code for the arm-gripper assembly, work surface, starting positions, and controls that move the arm and open or close the gripper.
Step 1: Download, Validate, and Inspect the Assets
Before any motion test, install Isaac Sim 6.2 and check its system requirements. The workflow also needs an agent that can use Isaac Sim skills, configured by following the Agent Skills documentation. Install the Hugging Face CLI only if you choose the command-line download option. The source also mentions using the USD Search endpoint, described as the SimReady Central Space search service endpoint, to look for manufacturer-curated content.
Open SimReady Central and find the UR17G arm, Robotiq 2F-140 gripper, and beverage can used in the example. Select Download on each asset and use Save to folder or the supplied Hugging Face command. Keep all three asset folders intact so Isaac Sim can find supporting files. Use the asset versions linked in the example, then give the agent the local USD file paths. The exact path depends on the download location.
Validation reports show what each asset was checked for. The arm report uses the Robot-Body 4.0.0 profile, and the gripper report uses Robot-Gripper 3.0.0. Both record passing PhysX driven-joint checks, which inspect joint-drive settings. This gives a shared baseline for reuse, but it does not prove the full assembled task will succeed.
Before running pick-and-place, test three things: joint settings, movement, and contact. Review the arm and gripper joint limits and drive settings. Test a small arm movement and open and close the gripper. Confirm the beverage can rests on the work surface without falling through it. During the task, watch whether the can stays in the gripper as the arm moves. Asset structures vary, so the loaded scene's settings are the starting point.
Step 2: Assemble the Robot and Compose the Scene
With the assets downloaded, start Isaac Sim from its installation directory with the remote Python server enabled. That allows the agent to interact with the simulation through Python. If using an agent, start it from the Isaac Sim directory so it can access the bundled skills needed for the workflow.
Confirm that all three assets appear without missing-file errors. Then inspect the arm and gripper's joint limits and drive settings, either through the agent or manually in Isaac Sim. The robot and gripper contain physics joints that can be tested and validated in the Isaac Sim graphical interface.
The next goal is to assemble the arm and gripper so they can be controlled as one articulation. A manual path is the Robot Assembler tutorial. In the agent-assisted example, the prompt asks the agent to complete assembly and validate the robot's physics. The agent ends up using the isaac-sim-remote and usd-articulation skills packaged with Isaac Sim. After that step, the scene has the assembled robot and the beverage can needed for the task.
The example uses a frontier AI model, GPT-6 Sol, to help create and run task code in an open Isaac Sim scene. The source notes that another agent and model combination may work if it can load the required skills and execute Python against Isaac Sim. Generated code and results can vary, so inspect the generated code and check the resulting behavior before adapting the task.
Step 3: Test Controls and Run the Pick-and-Place Sequence
With the robot assembled, the controller coordinates the UR17G arm's movement with the Robotiq 2F-140 gripper's opening and closing. The full task tests whether the gripper holds the can during movement and releases it at the target. Reset the scene, then run the approach, grasp, lift, hold, transfer, and release sequence. The example scene moves the beverage can from the yellow pickup marker to the green target. Compare the run with the opening video and the success criteria shown in the source.
The agent uses isaac-sim-remote, manipulaiton-ik, and physics-simulation skills for this part. If a run differs, note where the sequence fails and ask the agent to inspect the scene and generate controls. Reset the can and robot to the same starting state before retrying so results can be compared. To understand or adapt the controller, the source points to the Adding a Manipulator Robot tutorial.
The source is clear about limits. This is a fixed setup: the can starts at a known position, the robot is anchored, and the workspace is clear and flat. No vision was used, and collision checks were limited. The beverage can is simplified: deformation and liquid motion are not modeled, so the example does not test crushing.
The grasp and motion are scripted. A horizontal grasp uses a 2 N·m drive-effort cap chosen through simulation tests, not calibrated to fingertip forces or recommended by the manufacturer. Contact and friction hold the beverage can as the arm follows fixed waypoints without replanning. For hardware gripping-force specifications, the source points to the Robotiq product sheet.
For readers, the result shows a practical path from validated OpenUSD assets to a working simulation demo. It also shows what such a demo should not be used to claim: robustness to uncertain object positions, different payloads, or real hardware. The workflow can be reused to build other pick-and-place tasks, browse assets on SimReady Central, check profiles and validation reports, follow Isaac Sim installation guides, connect an agent, assemble robots manually or with skills, and prepare new SimReady assets using the linked 5 Steps to Create SimReady Assets for Robotics guidance. The source also points to Michael Wagner's voice-controlled virtual reality demo for inspiration and invites sharing scenes in Hugging Face comments, including assets and Isaac Sim version used.
Common mistakes
- Treating profile validation as proof of real-world performance. Passing Robot-Body 4.0.0 or Robot-Gripper 3.0.0 checks means the asset has the expected authored settings; it does not prove robust grasping on hardware.
- Downloading incomplete asset folders. If supporting files are moved or removed, Isaac Sim may not find the local USD paths.
- Starting the agent outside the Isaac Sim directory. The agent needs access to bundled skills, and the remote Python server must be enabled for Python interaction.
- Skipping pre-run checks. A stationary scene can look correct while joint limits, drive settings, movement, or contact are still wrong.
- Changing the starting state between retries. Resetting the can and robot to the same positions is necessary to compare results fairly.
- Assuming the beverage can behaves like a physical can. This simulation omits deformation and liquid motion, and the 2 N·m drive-effort cap was chosen through simulation tests, not calibrated to fingertip forces.
- Expecting every agent or model to produce the same code. The workflow works with GPT-6 Sol in the example, but generated code and results can vary with another agent and model combination.

Comments (0)