Ant Lingbo Tech Open-Sources Next-Gen Embodied AI Model LingBot-VLA 2.0
On July 8th, Ant Lingbo Technology announced the upgrade and open-sourcing of its new-generation embodied foundation model, LingBot-VLA 2.0. This represents a comprehensive enhancement over the initial LingBot-VLA 1.0, which was released in January. The upgraded model incorporates 60,000 hours of high-quality, real-world physical data during its pre-training phase. This extensive dataset covers 20 different robot configurations from 17 major robot brands. Furthermore, LingBot-VLA 2.0 expands its support for various degrees of freedom, including those for the head, waist, end-effectors, and mobile bases. The company states that these advancements result in significant improvements in configuration generalization, degree-of-freedom support, and deployment efficiency.
The open-sourcing of LingBot-VLA 2.0 by Ant Lingbo Technology signifies a strategic move to accelerate development and adoption within the embodied AI sector. By integrating a substantial volume of real-world physical data and broadening support for diverse robot hardware, the model aims to enhance its practical applicability. This approach democratizes access to advanced AI capabilities, potentially fostering a more competitive ecosystem and driving innovation across various robotics applications. The focus on generalization and deployment efficiency suggests an intent to bridge the gap between research and real-world implementation, a critical step in realizing the potential of AI in physical systems over the next decade.
AI-generated to prompt reflection — not editorial opinion, not advice, not a statement of fact. How this works.