What problem does it solve?
Running a robot live through an Isaac Sim scene in your own script involves hidden pitfalls: physics that never steps because the timeline is not playing, GPU out-of-memory crashes on 98K-prim stages, and Kit 110 hangs after kill-and-relaunch cycles. This Skill provides the runtime patterns, failure-mode fixes, and API references needed to navigate a robot reliably.
Core Features & Use Cases
- RL Policy Execution: Load and run locomotion policies via
PolicyController and drive articulations with isaacsim.core.experimental.prims.Articulation.
- Heavy-Stage Strategies: Choose between physics simulation, baked timeSamples, or per-frame transforms, and strip non-robot physics to keep 33K+ prim stages responsive.
- Kit 110 Gotcha Reference: Covers timeline.play requirements, asyncio event-loop subscription,
capture_viewport_to_file result handling, and GPU OOM recovery.
- Use Case: You have a Spot robot USD and a large warehouse stage. Use this Skill to load an RL policy, follow an A* trajectory from
navigation-primitives, and render a chase-camera run without hitting Vulkan OOM errors.
Quick Start
Ask the AI to write an Isaac Sim script that loads a Spot robot, plays the timeline, runs a PolicyController along a planned route, and avoids GPU OOM on a heavy stage.
