Galbot has integrated Nvidia’s Jetson AGX Thor platform into its G1 Premium humanoid robot, aiming to accelerate performance and expand commercial deployment across retail, healthcare, and logistics. The company claims it is the “first in the world” to use the new Nvidia chip tech in a humanoid robot.
Unveiled at the World Robotics Conference (WRC), the G1 Premium now delivers 7.5x the AI compute of Nvidia Jetson Orin and 3.5x greater energy efficiency. The upgrades enable faster, smoother motion, real-time reasoning, and higher-precision autonomy.
Professor Wang He, founder and CTO of Galbot, says: “Our G1 Premium, now running on Nvidia Jetson Thor, has already demonstrated remarkable advancements in speed and improved real-time reasoning capability. This early adoption empowers us to push our proprietary VLA models to new levels of real-world capability.”
At WRC, the robot showcased fluid navigation and precision drifting, earning recognition as the “swiftest humanoid robot worker”.
Wins gold at World Humanoid Robot Games
Galbot’s robot also secured first place at the World Humanoid Robot Games, competing fully autonomously without teleoperation. It completed the final challenge in 10 minutes and 22 seconds, scoring 336 points – more than 160 points ahead of the runner-up. The event highlighted its ability to execute complex tasks such as retrieving medications from shelves, underscoring potential for healthcare use cases.
AI foundation models and simulation
Galbot’s progress is built on its Sim2Real approach, using Nvidia Isaac Sim and Isaac Lab to pre-train robots on massive synthetic datasets before refining with minimal real-world data. The company says this method reduces training costs while improving generalizability in complex environments.
On this foundation, Galbot has developed proprietary embodied AI foundation models – GraspVLA, TrackVLA, and GroceryVLA – supporting zero-shot grasping, navigation, and dense retail operations. The company has also created tens of billions of synthetic data points for grasping and deformable object manipulation, enabling greater robustness in real-world conditions.
OpenWBT_Isaac platform
In collaboration with Tsinghua University and Shanghai Qi Zhi Institute, Galbot has launched OpenWBT_Isaac, a simulation platform for whole-body teleoperation of humanoid robots. Built on Nvidia Isaac tools and accelerated by L20/RTX 5880 GPUs, it supports rapid deployment, cross-platform compatibility, and high-efficiency data collection.
Real-world deployment
Galbot says its G1 robots are already operating in more than 10 smart pharmacies in Beijing, with plans to expand to 100 by year-end. The company is also working with Bosch Group to bring embodied AI-powered robotics into smart manufacturing, replacing conventional programmed arms with adaptable systems.
