Meta open sources Project SuperDex for robot dexterity
Project SuperDex combines contact focused simulation, headset teleoperation and data generation, with a claimed zero shot transfer to a real robotic hand.
-
-
Unitree G1 teardown reveals actuator and cooling tradeoffs
Munro Live’s full teardown of the Unitree G1 shows planetary actuators, unusual heat pipe cooling, extensive CNC machining and production compromises.
-
BeyondMimic teaches Unitree G1 humanoid broad motion set
BeyondMimic used 2.5 hours of human motion data to train a Unitree G1, then transferred 30 representative motions to hardware without task specific tuning.
-
Musashi humanoid demonstrates low speed car driving
University of Tokyo’s Musashi humanoid operated a regular car’s controls and recognized traffic signals, although its early speed was limited to 3 mph.
-
GMO unveils humanoid ambulance for on site robot repairs
GMO Air’s mobile workshop carries engineers, repair gear and a spare humanoid, letting operators replace damaged robots while repairs continue off site.
-
Figure says robotics dataset reaches 86,000 weekly users
Figure says 86,000 weekly active users are uploading data to its robotics dataset, with a live map visualizing contributions around the world.
-
JoyIn unveils Aether humanoid model amid OpenAI dispute
JoyIn says its Aether control model cuts humanoid training time and improves first attempt task success, but its OpenAI allegations remain unverified.
-
NVIDIA trains SONIC for general humanoid motion control
NVIDIA’s SONIC controller learned from more than 100 million human motion frames and accepts VR, video and language based commands without retraining.
-
Tiangong Ultra targets practical work after 8.64 second sprint
X-Humanoid is prioritizing reliability and practical tasks after its 75 kg Tiangong Ultra humanoid completed a 100 meter sprint in 8.64 seconds.
-
Schaeffler trains Hexagon’s AEON humanoid for factory work
Hexagon Robotics’ AEON has entered Schaeffler’s German Humanoid Gym for task training and validation ahead of planned global factory deployments.
