Daily Digest
Daily brief
Loading brief…
Issues & signals
- [Bug Report] RigidObjectCollection deletion callback treats an event as a pathgithub:isaac-sim/IsaacLab9/14/2026crashes-stabilitycrashhardwareisaac-simisaac-lab
### Describe the bug The PhysX `RigidObjectCollection._on_prim_deletion()` override treats its callback argument as a path string. The callback registration in `AssetBase` passes the PhysX event object instead, with the deleted path stored in `event.payload["
- [Bug Report] Camera intrinsic batch mismatch silently updates only a prefixgithub:isaac-sim/IsaacLab9/14/2026asset-pipelineusdhardwaresensorsisaac-simisaac-lab
### Describe the bug `Camera.set_intrinsic_matrices()` accepts a batch of matrices and a selection of camera IDs, but it does not require the two counts to match. The write loop uses `zip(env_ids, matrices)`, which silently stops at the shorter input. With t
- Studio web viewer is unusable: studio.web.headless_ui is missing from the wheelgithub:google-deepmind/mujoco9/14/2026renderingrenderinghardwaresensorsdocsmujoco
### Intro Hi! I'm from AWS. I use MuJoCo for headless physics simulation and offscreen rendering distributed across many workers, and I have been trying Studio's Python bindings to add an in-viewer UI. ### My setup Reproduced on both: - MuJoCo 3.13.0 (PyP
- Studio native viewer cannot start on macOS: NSWindow created off the main threadgithub:google-deepmind/mujoco9/14/2026renderingrenderingsensorsdocsmujoco
### Intro Hi! I use MuJoCo for headless physics simulation and offscreen rendering distributed across many workers, and I have been scripting the viewer to drive job submission. That is what led me into experimental/studio. ### My setup - MuJoCo 3.13.0, in
- [BUG] KAMINO two-dimensional sparse GEMV paths cannot launchgithub:newton-physics/newton9/14/2026renderingrenderinghardwarenewtonwarp
### Bug Description KAMINO's generalized block-sparse matrix-vector functions advertise both flattened 1D vectors and 2D `(num_matrices, max_dimension)` batches, but both 2D branches are miswired. `block_sparse_gemv` selects `_make_scale_vector_kernel_2d` bu
- Failed cudaStreamEndCapture skips clean_up(), disabling graph capture for the processgithub:NVIDIA/warp9/14/2026hardware-integrationhardwarewarp
### Summary `wp_cuda_graph_end_capture()` in `warp/native/warp.cu` defines a `clean_up()` lambda that unwinds the capture bookkeeping: the `g_captures` entry, the graph allocation table, and the terminating `EndCapture`. Every failure return below that defini
- [Bug Report] DirectMARLEnv state_space=0 is exposed as Box(0) instead of Nonegithub:isaac-sim/IsaacLab9/14/2026training-infrarlrenderinghardwaremulti-agentenv-apidocs
### Describe the bug `DirectMARLEnvCfg.state_space = 0` is documented to disable centralized state and make `env.state_space` equal to `None`. Instead, `DirectMARLEnv._configure_env_spaces()` assigns `None` and then immediately overwrites it with `gym.spaces.
- [Bug Report] TerminationManager.reset reports statistics from environments outside env_idsgithub:isaac-sim/IsaacLab9/14/2026renderingrenderinghardwaredeploymentdocsisaac-simisaac-lab
### Describe the bug `TerminationManager.reset(env_ids)` reports each termination term's mean over every environment, even when only a subset is being reset. The method documents that `env_ids` selects the environments considered, but `_last_episode_dones` is
- 为什么现在需要ros?github:isaac-sim/IsaacLab9/15/2026deploymentdeploymentperceptionintegrationisaac-sim
Checking to see if RMW can be loaded: failed to get symbol 'rmw_init_options_init' due to Environment variable 'AMENT_PREFIX_PATH' is not set or empty, at /workspace/humble_ws/src/rmw_implementation/src/functions.cpp:171, at /workspace/humble_ws/src/rcl/rcl/sr
- mj_recompile reads past saved actuator state when an actuator's actdim increasesgithub:google-deepmind/mujoco9/15/2026training-infrarlusddeploymentdocsmujoco
### Intro Hi! I am testing state continuity across live `mjSpec` schema changes. I first sent this finding to Google's vulnerability intake, and they advised reporting it as an issue in the MuJoCo repository. ### My setup - MuJoCo 3.13.0 at `123347c0eeab7
- [Jetson AGX Thor] devfreq_wq deadlock with rmapiLockAcquire causes GPU tasks to hang in D state ( JetPack 7.2)nvidia-forum:robotics-edge-computing9/14/2026crashes-stabilitycrashhardware
- mj_recompile reads old joint state using the new joint-type widthgithub:google-deepmind/mujoco9/14/2026training-infrarlusddeploymentdocsmujoco
### Intro Hi! I am testing state continuity across live `mjSpec` schema changes. Changing an existing joint from hinge to free before `mj_recompile` makes state preservation read the old `mjData` with the new joint width. ### My setup - MuJoCo source buil
- [Model editing] Failed allocations raise C++ exceptions through the C APIgithub:google-deepmind/mujoco9/14/2026deploymentdeploymentdocsmujoco
### Intro Hi! I am a researcher at University of Ljubljana, I use MuJoCo for different things. ### My setup MuJoCo 3.13.0; C API (C++ code), x84, Fedora 44. ### What's happening? What did you expect? Model editing functions, such as [`mjs_addBody`](https
- Studio: an exception in one plugin handler terminates the viewergithub:google-deepmind/mujoco9/14/2026renderingrenderingsensorsdocsintegrationmujoco
### Intro I use MuJoCo for headless physics simulation and offscreen rendering distributed across many workers, and I have been writing a Studio plugin that adds a panel to the viewer. ### My setup Observed on both: - MuJoCo 3.13.0 (PyPI wheel), Python 3.1
- [BUG] KAMINO sparse diagonal preconditioner fails while resolving its dtypegithub:newton-physics/newton9/14/2026renderingrenderinghardwarenewtonwarp
### Bug Description KAMINO's sparse `jacobi_diagonal` preconditioner cannot construct its elementwise inverse kernel. `_make_cwise_inverse_kernel_2d(dtype)` uses `dtype` in the `wp.array2d[dtype]` annotation, but nowhere in executable code. With postponed ann
- [REQ] Allow known dependency deprecations in strict CI jobsgithub:newton-physics/newton9/14/2026training-infrarldxnewtonwarp
## Description Newton's strict warning checks need a way to acknowledge known dependency deprecations in jobs that test newer or nightly dependency versions. An acknowledged deprecation should remain visible in the test output without failing the job. New or
- [BUG] KAMINO scalar sparse blocks fail BLAS kernel code generationgithub:newton-physics/newton9/14/2026renderingrenderinghardwarenewtonwarp
### Bug Description `BlockDType(dtype=wp.float32)` is the documented/default representation for scalar sparse blocks, but KAMINO's generated sparse BLAS kernels index each scalar value as `block[j]` or `block[i, j]`. Warp therefore fails kernel code generatio
- [Bug Report] DelayBuffer.set_time_lag leaves invalid state after raising ValueErrorgithub:isaac-sim/IsaacLab9/14/2026hardware-integrationhardwaredeploymentisaac-simisaac-lab
### Describe the bug `DelayBuffer.set_time_lag()` validates the requested lags only after writing them into `_time_lags` and recomputing `_min_time_lag` / `_max_time_lag`. When the method raises for a negative lag or a lag above `history_length`, the rejected
- `smoothnormal="false"` is silently ignored for meshes loaded via a decoder plugin (e.g. STL)github:google-deepmind/mujoco9/14/2026asset-pipelineusdrenderingdeploymentdocsintegrationmujoco
### Intro Hi! I am a engineer at Boston Dynamics, I use MuJoCo for making dancing robots. ### Summary `<mesh smoothnormal="false">` (the default) is documented as excluding large-angle faces from vertex-normal averaging, so hard edges stay sharp. For meshe
- Qwen3.8-Flash-Next at 46.7 tok/s on Jetson AGX Thornvidia-forum:robotics-edge-computing9/14/2026▲ 1hardware-integrationhardware
- Kernel Panic in Soft Interrupt Context with JetPack 6.0nvidia-forum:robotics-edge-computing9/15/2026other
- AGX Orin J40 USB port always recognized as USB2.0nvidia-forum:robotics-edge-computing9/15/2026hardware-integrationhardware
- Tilted camera causes chaotic vo_pose when enable_imu_fusion=Truenvidia-forum:isaac9/15/2026sensors-perceptionsensors
- Tilted camera causes chaotic vo_pose when enable_imu_fusion=Truenvidia-forum:isaac-ros9/15/2026sensors-perceptionsensors
- Tilted camera causes chaotic vo_pose when enable_imu_fusion=Truenvidia-forum:robotics-edge-computing9/15/2026sensors-perceptionsensors
- Agx orin ssh无法登录nvidia-forum:robotics-edge-computing9/15/2026hardware-integrationhardware
- Null pointer derefence on DisplayPort IRQ in JetPack 7 display drivernvidia-forum:robotics-edge-computing9/15/2026hardware-integrationhardware
- Jetson Orin JetPack 7.2 custom board: Type-C reverse insertion issue with TUSB321AI PD controller and USB Composite Device enumerationnvidia-forum:robotics-edge-computing9/15/2026hardware-integrationhardware
- Official implementation steps for the Thor T5000 stress testnvidia-forum:robotics-edge-computing9/15/2026hardware-integrationhardware
- How to implement dual display (DP + HDMI) on Jetson AGX Orin given shared pins?nvidia-forum:robotics-edge-computing9/15/2026hardware-integrationhardware
- Serial-number 这个节点怎么开启nvidia-forum:robotics-edge-computing9/15/2026other
- AGX Orin C5 NVHS PCIe RC-EP tvnet: R36.4.3 stuck Polling.Compliance (0x2018); works on JP5.1.3nvidia-forum:robotics-edge-computing9/15/2026hardware-integrationhardware
- Jetson Orin NX does not boot up in super modenvidia-forum:robotics-edge-computing9/15/2026hardware-integrationhardware
- Add a fluid checkpoint example with a custom backward passgithub:NVIDIA/warp9/15/2026hardware-integrationhardwarewarp
### Description Add `warp/examples/optim/example_fluid_checkpoint_custom_backward.py` alongside the existing fluid checkpoint example. `example_fluid_checkpoint.py` shows how to checkpoint a simulation across timesteps, but the pressure solver still records
- Implementing multi-rate Ethernet (1G/2.5G/10G) with Jetson AGX Orin MGBEnvidia-forum:robotics-edge-computing9/15/2026hardware-integrationhardware
- AGX Thor add second fan on custom boardnvidia-forum:robotics-edge-computing9/14/2026hardware-integrationhardware
- mj_recompile resets documented integration-state fields on a no-op recompilegithub:google-deepmind/mujoco9/14/2026training-infrarldeploymentdocsintegrationmujoco
### Intro Hi! I am testing state continuity across live `mjSpec` recompilation. I found that a no-op `mj_recompile` preserves a subset of `mjData` but resets other fields that MuJoCo documents as integration state. ### My setup - MuJoCo source build from
- Orin nx, sdk:7.2.1, ubuntu24.04, es8389配置nvidia-forum:robotics-edge-computing9/14/2026hardware-integrationhardware
- [Bug] USD fallback ignores `resetXformStack` when reading world posesgithub:isaac-sim/IsaacSim9/14/2026asset-pipelineusdrenderinghardwareintegrationisaac-simnewton
### Description ## Summary The experimental C++ prim reader returns incorrect world poses when the USD fallback is used on prims with `resetXformStack`. A transform reset should stop the prim from inheriting its parent's transform. The reader still includes
- GCC version mismatch warning when building out-of-tree kernel modules on Jetson R39.2.1 (13.2.0 vs 13.3.0)nvidia-forum:robotics-edge-computing9/14/2026hardware-integrationhardware
- 在Jetpack7.2的环境下,Orin Nx模组能否通过U盘对QSPI或者NVME单独刷机nvidia-forum:robotics-edge-computing9/14/2026hardware-integrationhardware
- Isaac Sim 6.1 General Availabilitynvidia-forum:simulation9/14/2026otherv6.1isaac-sim
- AGX Orin Devkit bricked after QSPI capsule update from L4T 35.4.1 to 35.5.0 — black screen, APX mode accessiblenvidia-forum:robotics-edge-computing9/14/2026hardware-integrationhardware
- Where are the sources of spidev_test modified by nvidia?nvidia-forum:robotics-edge-computing9/14/2026other
- [BUG] KAMINO sparse metadata views can bind CUDA pointers as CPU arraysgithub:newton-physics/newton9/14/2026renderingrenderinghardwarenewtonwarp
### Bug Description KAMINO constructs several raw-pointer array views without fully preserving the backing allocation's device and lifetime. In `BlockSparseMatrices`, `max_rows` and `max_cols` omit `device`, so Warp assigns the current ambient device when the
- Jetson Orin Nano R36.4.3 Browser Installationnvidia-forum:robotics-edge-computing9/14/2026hardware-integrationhardware
- Trouble starting docker container with cuda after enabling MIGnvidia-forum:robotics-edge-computing9/14/2026hardware-integrationhardware
- [Bug Report] EventManager.apply(reset, env_ids=None) passes slice(None) to event termsgithub:isaac-sim/IsaacLab9/14/2026renderingrenderinghardwaredeploymentisaac-simisaac-lab
### Describe the bug `EventManager.apply(mode="reset", env_ids=None, ...)` changes the documented `None` all-environment selector into `slice(None)` for its internal trigger bookkeeping, then passes that slice to the reset event callback. This violates the e
- DevKit Idle SoC Temperaturenvidia-forum:robotics-edge-computing9/14/2026other
- mjUI rendering corruption on Windows with recent AMD graphics drivergithub:google-deepmind/mujoco9/14/2026asset-pipelineusdrenderinghardwaredocsmujoco
### Intro Hi, I encountered this issue while using MuJoCo for a robotics simulation project on Windows. ### My setup - MuJoCo: 3.13.0 - API: Python - Python: 3.13.15 - glfw: 2.10.2 - OS: Windows 11 64-bit - Laptop: Lenovo IdeaPad 3 15ABA7 - GPU: AMD Radeon
- AGX THOR T5000 Mgbe QSFP28 AC caps position and voltagenvidia-forum:robotics-edge-computing9/14/2026hardware-integrationhardware
- [IGX Thor T7000] BMC remains in “Starting” state and never finishes initializationnvidia-forum:robotics-edge-computing9/14/2026hardware-integrationhardware
- Request for official method to validate 100 INT8 TOPSnvidia-forum:robotics-edge-computing9/15/2026feature-requestsfeature-request