Phynetic AI Weeklyby Phynetic
Section
23 items · newest first

Data, Sim & Tools

#
17 Sep

Isaac Lab 3.0 Early Access puts one task API across physics and rendering backends (tool · 8 minute read)

Isaac Lab 3.0 Early Access brings one task API across multiple physics, rendering and visualization backends, so supported tasks can switch between them without rewriting the environment. Many workflows now run kit-less, without installing or launching Isaac Sim, and a new isaaclab command replaces library-specific training and playback scripts. Quaternions change from WXYZ to XYZW, and General Availability is targeted toward the end of October 2026.

— GitHubopen-source-toolssimulation
#
14 Sep

FoldNet++: a Large-Scale Synthetic Dataset for Robotic T-Shirt Folding and Unfolding (paper · 15 minute read)

FoldNet++ spans 6 robotic embodiments, 1K T-shirts, 1K environmental assets and 120K episodes with rich annotations. It generates physically simulatable T-shirts with annotated semantic keypoints, from which a unified rule-based framework produces demonstrations for each embodiment. The authors report that visuomotor policies trained solely on this synthetic data achieve over 90% end-to-end task success when deployed directly to unseen real-world environments and previously unseen T-shirts.

— cs.RO updates on arXiv.orgdatasets-benchmarksmanipulation
#
9 Sep

MuJoCo 3.13 adds a discrete integrator that keeps stiff springs stable (tool · 2 minute read)

MuJoCo 3.13.0 writes the .mjz root file as model.xml, which survives file renaming, and lets site geometries be associated with meshes. A new discrete integrator merges the constraint solve and the implicit velocity update into one operation, bringing joint, tendon and actuator stiffness into the solver's metric. Passive springs and actuator position gains stay stable at timesteps far beyond the explicit stability limit, and Python 3.15 is now supported, GIL and free-threading alike.

— GitHubopen-source-toolssimulation
#
8 Sep

Robotics with Conan: consuming ROS as a regular package (4 minute read)

Conan's team shows how to consume ROS as an ordinary Conan package, rather than installing it system-wide with apt or rosdep on every developer machine and CI agent. It addresses the usual split where C++ control and perception code is managed with Conan while ROS sits separately on top of it. Conan's team has been building the Kilted ROS distribution from source as a regular package for Linux, macOS and Windows, so adding ROS to a C++ project becomes one more requires.

— blog.conan.ioopen-source-tools
#
7 Sep

Genesis 1.4 exports scenes and trajectories for bit-exact replay (tool · 3 minute read)

Genesis v1.4.0 can export a scene and a complete trajectory as a standalone archive that replays bit-exactly on any machine, aimed squarely at making bug reports reproducible. Only the rigid solver is supported so far, and the static description of entities is now separated from the runtime getters. The breaking changes cover external wrenches applied at any point on a link and morph pose offsets in the velocity and acceleration getters, alongside a long list of asset-parsing fixes.

— GitHubopen-source-toolssimulation
#
2 Sep

AGIBOT Releases AGIBOT WORLD 2026 Theme 3 for Real-World Robot Reinforcement Learning (4 minute read)

AGIBOT has open-sourced 11,430 real-world trajectories across 14 industrial and household tasks, including 1,024 successful and 1,369 failed policy rollouts. Annotations cover task progress, completion, errors, disturbances and human interventions, giving structured execution feedback rather than demonstrations alone. It is the reinforcement-learning theme of AGIBOT WORLD 2026, an open-source dataset initiative organised into themes, and the data spans expert demonstrations, autonomous deployment and human-in-the-loop correction.

— AgiBot newsopen-source-toolsdatasets-benchmarksteleoperation-data-collection
#
1 Sep

Show HN: I kept finding robot datasets with skewed clocks, so I built a checker (tool · 9 minute read)

embodied-sync is an open-source Python library for aligning and validating multi-sensor robot datasets, handling different sample rates, clock skew and drift, late packets and missing frames. It fits around UMI, LeRobot, ROS 2 and MCAP, and the author describes the project as alpha. The pitch is that cameras run at one rate and robot state at another, packets arrive late, and a device reconnect can silently reset its clock, so a policy trains on observations that never lined up.

— GitHubdatasets-benchmarksopen-source-tools
#
31 Aug

Hebbian Robotics open-sources HFlow, an SDK for robot data pipelines (tool · 7 minute read)

Hebbian Robotics, part of Y Combinator's summer 2026 batch, has open-sourced HFlow, an SDK for multimodal data pipelines in robotics. It checks recordings for frozen cameras, drifting streams, missing topics and duplicates, runs transformations and labels, and stamps each processed episode with its provenance in a queryable catalogue, with MCAP as its input and output format.

— GitHubopen-source-toolsdatasets-benchmarks
#
27 Aug

Hugging Face is selling a cute $399 open source duck robot, Microduck (2 minute read)

Hugging Face unveiled the Microduck, a $399 open-source robot that ships before Christmas. The 25-centimetre duck waddles, picks things up with its beak, rights itself after a fall and roller skates, following the Reachy Mini in a hardware line the company started by acquiring Pollen Robotics. It perceives with a camera, lidar and two inertial measurement units.

— Robotics News | TechCrunchopen-source-toolshardware
#
25 Aug

Introducing Index: Building The World’s Largest and Most Diverse Physical Dataset (3 minute read)

Figure has come out of stealth with Index, a robot training dataset it says is the most diverse built so far. In four months its creator network uploaded over 16 million videos from more than 100 countries, processing 30 minutes of video every second, with $15 million paid out to contributors. Figure crossed 264,000 app downloads and 44,000 weekly active users while still in stealth, and says it will spend over $1 billion on data and compute in the next twelve months.

— Figure newsdatasets-benchmarksteleoperation-data-collectionvla-foundation-models
#
20 Aug

MuJoCo 3.12 moves MJCF into a single schema and adds a PID actuator (tool · 6 minute read)

MuJoCo 3.12.0 defines the MJCF grammar in a single schema file, generating the parser's grammar table, keyword maps and typed attribute bindings from it. The release also adds a PID actuator and, as a breaking change, drops the custom binary texture format in favour of PNG and KTX. The new PID actuator takes real position and velocity setpoints with optional integral action, anti-windup and slew limiting, subsuming the old plugin with proper activation state.

— GitHubopen-source-toolssimulation
#
20 Aug

Keeping the Franka Emika Panda alive: a ROS 2 stack with a reliable position interface (paper · 15 minute read)

The Franka Emika Panda's external position interface has been unreliable for years, and this paper traces the vibrations and protective stops to the timing of the external control loop and sampling jitter rather than to the robot. The open-source stack decouples real-time communication from the ROS 2 control loop and adds rate matching and position-domain reference generation. It is validated across motion planning, compliance control, position-controlled manipulation and haptic teleoperation, on two independent Panda platforms.

— arXivopen-source-toolshardwaremanipulation
#
14 Aug

HumanTracker: Galaxy General Robotics' humanoid motion-tracking benchmark (dataset · 1 minute read)

Galaxy General Robotics' HumanTracker benchmark scores humanoid motion tracking against 2,500 reference clips retargeted to a 29-DoF Unitree G1-style body, in four families from daily motion to highly dynamic moves. It adds 6,000 human preference pairs comparing rollouts from four trackers — GMT, TWIST2, SONIC and Humanoid-GPT — which train a HumanScore reward model meant to catch foot skating and unstable support that per-frame error misses. It is Apache 2.0, with an ECCV 2026 paper.

— Hugging Facedatasets-benchmarkshumanoids
#
12 Aug

ZEST turns motion capture, video and animation into robot skills without fine-tuning (3 minute read)

Researchers at the RAI Institute and Boston Dynamics describe ZEST, a reinforcement-learning recipe that turns motion capture, single-camera video and keyframe animation into whole-body skills, trained entirely in simulation and run on hardware without fine-tuning. On Atlas, Unitree's G1 and Spot it produced crawls, cartwheels, breakdancing and a backflip, each policy taking about 10 hours on one NVIDIA L4 GPU; so far it is limited to flat, non-slippery ground.

— interestingengineering.comhumanoidslocomotionsimulation
#
12 Aug

NVIDIA JetPack 7.2.1 Adds Agentic Video Skills and T3000 Emulation (4 minute read)

NVIDIA JetPack 7.2.1 adds PyNvVideoCodec 2.2 on Jetson Thor, bringing the hardware video encoder and decoder that JetPack 7.1 opened to C and C++ developers to Python for the first time. NVIDIA calls video a core data path in Jetson applications from robotics to industrial automation, where the engineering lies in choosing the interface, codec, pixel format and memory path, and proving the pipeline meets its latency and throughput goals.

— Robotics – NVIDIA Technical Bloghardware
#
11 Aug

TLabel – A unified annotation format and converter for tactile robotics datasets (tool · 3 minute read)

TLabel is a cross-sensor annotation schema for tactile manipulation data, producing 14-dimensional semantic annotations that different sensor types can all emit. Each adapter declares which dimensions it can and cannot annotate, so results from different tactile sensors become comparable rather than sensor-specific. Compliance levels L1 to L4 let each sensor participate at whatever information density it can actually support.

— GitHubmanipulationhardwareopen-source-tools
#
7 Aug

Genesis 1.3.2 makes rigid-body simulation deterministic on CPU and GPU (tool · 1 minute read)

Genesis v1.3.2 makes rigid-body simulation behave consistently across scene orientations and scales, and improves numerical robustness when the mass matrix is ill-conditioned. Simulation is now fully deterministic on a given machine, on both CPU and GPU. It also adds terrain-height querying, and fixes viewer picking, raycast sensors missing hits, and a crash on delayed sensor reads.

— GitHubopen-source-toolssimulation
#
7 Aug

Video to Data: NVIDIA's robot dexterity task library and dataset (dataset · 1 minute read)

NVIDIA has released 1,215 human manipulation demonstrations from the TACO dataset, retargeted to bimanual Sharpa hands, with 121,500 robot episodes generated from them by its Video to Data pipeline. The 924.8 GB set, in Parquet and LeRobot v3 format under CC-BY-4.0, is meant to show what the pipeline outputs and to let users try its grounding step or downstream training.

— Hugging Facemanipulationdatasets-benchmarks
#
3 Aug

LeRobot v0.6.1 renames a core module and loads safetensors straight to the device (tool · 7 minute read)

LeRobot v0.6.1 renames the lerobot.types module to lerobot.lerobot_types, a breaking change from v0.6.0. The release also loads safetensors directly onto the target device and improves the vision-language subtask annotation tooling. Other changes resolve policy components by convention and aggregate policy sub-losses through the metrics tracker.

— GitHubopen-source-toolsvla-foundation-models