Galbot picks up $153M to commercialize G1 semi-humanoid

Galbot stated its G1 robotic can work in quite a lot of environments, together with pharmacies. | Supply: Galbot

Galaxy Common Robotic Co., also called Galbot, this week accomplished a funding spherical of RMB 1.1 billion, or round $153 million. The corporate stated this brings its whole funding to about $335 million raised up to now two years.

In March, Galbot launched its flagship robotic, the G1. The semi-humanoid cellular manipulator options wheels and two arms. It’s designed to automate stock, replenishment, supply, and packaging. The corporate stated its synthetic intelligence fashions allow G1 to deal with 5,000 several types of items.

Galbot stated the robotic solely takes a day to deploy into a brand new retailer. Presently, practically 10 shops in Beijing have deployed the robotic. The corporate expects to deploy it into 100 shops throughout the nation throughout the yr.

Based in Could 2023, Galbot makes use of simulated artificial motion knowledge units for pre-training and actual knowledge for post-training. The Beijing-based firm stated this strategy results in the fast iteration of embodied massive mannequin know-how worldwide.

VLA fashions energy G1 robotic

Galbot stated it has developed three proprietary vision-language-action fashions (VLAs). The primary, launched in January 2025, is GraspVLA, which the corporate described as an “end-to-end” embodied mannequin that’s pre-trained on simulated artificial knowledge.

Additionally, Galbot stated it’s working with NVIDIA Isaac Sim, and the mannequin can obtain zero-shot generalization with solely pre-training.

Subsequent, it created GroceryVLA, a mannequin for retail commercialization. Galbot stated this mannequin solves two key challenges in complicated retail environments:

• Clever greedy of many objects: In a situation involving complicated and densely stacked items, the mannequin doesn’t want to regulate parameters for every product individually, Galbot stated. Whether or not it’s tender baggage, onerous packing containers, bottles, glass containers, plastic packaging, or extra, the mannequin allows robots to precisely grasp the merchandise, the corporate claimed.
• Excessive effectivity in scene adaptation: Galbot stated the mannequin has sturdy autonomous decision-making capabilities, reminiscent of dynamically deciding on the optimum operation goal. It additionally has anti-interference capabilities, which means it might probably cope with unintended displacement or dumping of products. Extra importantly, it doesn’t require any scene pre-collecting to deploy.

Lastly, Galbot presents TrackVLA, a product-level navigation mannequin. The corporate stated it might probably carry out a spread of duties, together with pure visible atmosphere notion, comprehension of natural-language instructions, zero-sample generalization in complicated scenes, following extremely dynamic targets, and extra.

Galbot invests in collaboration

CATL and Puquan Capital led Galbot’s funding spherical. Different traders included China Improvement Financial institution, China Improvement Financial institution Science and Expertise Innovation Fund, Beijing Robotic Trade Fund, Qiming Enterprise Companions, and extra.

CATL was the core strategic investor. The businesses plan to collaborate on creating purposes for Galbot’s normal embodied intelligence massive mannequin and selling it.

Galbot has additionally labored to strengthen its ties to the automotive business. Final month, it introduced a three way partnership with Boyuan Capital, a subsidiary of the worldwide industrial large Bosch Group. The startup additionally signed a tripartite strategic memorandum with Bosch China and Boyuan Capital.

The companions stated they are going to deal with industrial manufacturing and collectively selling business purposes for embodied clever robots.