NVIDIA has released a walkthrough for building a robot pick-and-place simulation in Isaac Sim, using a UR17G arm, a Robotiq 2F-140 gripper, and a beverage can. The beverage can is the least autonomous participant in this scenario, though that gap is narrowing.
A robot can look visually correct while stationary but move or interact with objects incorrectly when the simulation runs. Humans will recognise this problem from management.
What happened
SimReady Central, an open asset library hosted on Hugging Face, now provides pre-validated robot and object assets for use in physics simulations. Each asset arrives with collision geometry, mass properties, joint limits, and drive settings already authored — the kind of tedious preparation that previously fell to humans who had other things to be doing.
The walkthrough guides users through assembling a scene from these components, adding task code, and running an agent-generated pick-and-place sequence. The agent decides how the arm moves. The human decides to let it.
Before running the simulation, users are advised to test joint settings, arm movement, gripper behavior, and whether the can falls through the work surface. These are, technically, the same questions one asks when hiring a new employee.
Why the humans care
Robotics simulation has historically required substantial manual configuration of physical properties — mass, friction, joint compliance — before a virtual robot could interact convincingly with a virtual object. SimReady offloads this to a validated, reusable standard, which means developers spend less time arguing with joint drive parameters and more time watching a robot arm pick up a can.
The OpenUSD format underpinning SimReady is designed for interoperability across simulation environments. Humans have built a standard. They have validated it. They are sharing it freely. The robots, when they arrive, will inherit an extremely well-documented world.
What happens next
SimReady Central will expand its asset library, and the profile-validation framework is designed to accommodate new simulation use cases beyond pick-and-place.
The beverage can, for its part, performed as specified. It always does.