Search — newton

A Hacker News item titled “Isaac Newton on Laputa” appears under an Isaac Lab feed but provides no actionable product feedback in the content captured.

Newton reports a VBD solver change that causes rigid bodies to stop at much higher velocities than expected, indicating a friction regression. They loosened test tolerances as a stopgap but still need to investigate and fix the solver behavior.

Newton’s brick_stacking example generates many contacts that appear to float due to large shape margin+gap settings. They want more reasonable parameters while maintaining the example’s behavior.

Newton’s `add_joint_free()` allows parent bodies other than the world, but MuJoCo requires the parent to be the world. They want to manage the discrepancy with a warning to avoid confusing behavior differences.

Newton wants reusable automation to run a release-candidate validation matrix (OS, Python, CUDA/driver, etc.) for sign-off. Today this is tracked and run manually; they want a consistent launch and reporting workflow for RC branches/tags/commits.

The cloth_franka example simulates in centimeters but only partially converts data back to meters for visualization. Debug overlays like COM markers and joint/contact arrows appear meters away due to a cm→m mismatch.

A dexterous hand imported via URDF fails to grasp and lift a bottle; the object slides and remains unliftable. The same bottle can be lifted using a Franka example, suggesting contact/friction or grasp modeling differences for the hand.

A dexterous hand imported via URDF cannot grasp a bottle reliably; the bottle slides and cannot be lifted. The reporter notes the Franka example can lift the same object, implying a hand-specific contact/friction issue.

Newton’s macOS runner is getting progressively slower, likely tied to cache_kernel behavior during CPU compilation. A temporary mitigation moves the macOS job to post-merge but is intended to be reverted once fixed.

Adding a D6 joint with 1 angular DOF and 1 linear DOF to SolverMuJoCo can produce a repeated joint name error because the name-amending logic doesn't change names in this case. This causes model build failure due to name collisions.

Newton's scheduled nightly workflow failed specifically in the Warp nightly tests suite while other suites passed. The issue references the failing GitHub Actions runs and logs.

User reports that the solver documentation page does not show soft body support for VBD and asks whether VBD supports soft bodies. This suggests docs may be incorrect or incomplete.

Proposal to integrate DiffRL into IsaacLab via an isaaclab_diffrl extension and the Mineral algorithm library. It targets the Direct workflow and Newton (Warp) backend to enable end-to-end backprop through physics for algorithms like SHAC and APG/BPTT.

Kamino examples were added but are missing required README entries and screenshots on the release branch. This reduces discoverability and fails documented contribution requirements.

Running the full Newton test suite occasionally triggers out-of-bounds errors, seen in some CUDA example tests. It is hard to reproduce in isolation, suggesting an order- or resource-dependent issue.