HomeArticle

Teslaium and Ubtech Reach Strategic Cooperation to Build Next-Generation Intelligent Agent Robots | Breaking News

黄 楠2025-09-17 21:23
Promote the in-depth integration of "spatial intelligence" and "embodied intelligence".

Author | Huang Nan

Editor | Yuan Silai

Recently, Telslian (a leading company in spatial intelligence) and Ubtech Robotics (the first publicly - traded company in humanoid robots) announced a strategic cooperation. The two sides will carry out in - depth cooperation around "spatial intelligence + embodied intelligence", open up and share innovative technologies in spatial intelligence and embodied intelligence, and jointly promote the full implementation of intelligent agent robots.

Currently, the global embodied intelligence industry is moving from the "laboratory demonstration" stage to the "initial industrial application" stage. The technology is continuously iterating, and there is high capital attention. However, large - scale commercialization still faces challenges such as technological bottlenecks, computing power costs, and scenario implementation.

Embodied intelligence is regarded as the combination of a "thinking brain" and an "actionable body". However, the complex, changeable, and unstructured scenarios in the real environment, such as light changes, object occlusion, and uneven ground, pose a severe test to its technological generalization ability.

To enable machines to truly "see" and "understand" the physical world and achieve real autonomous interaction with the physical world, spatial intelligence is the key.

Centering around the strategic goal of spatial intelligence, Telslian has split its business into AIoT infrastructure, AIoT domain models, and AIoT intelligent agents, corresponding to the three - fold capabilities of spatial intelligence respectively.

Among them, the AIoT domain model is the key to enabling machines to see and understand. At the domain model layer, Telslian's self - developed large spatial model breaks through the limitations of traditional language models, has multi - modal fusion and reasoning capabilities, can deeply understand and process physical laws including gravity, friction, and three - dimensional spatial relationships, and achieve the decomposition and planning of complex tasks in real scenarios.

At the AIoT infrastructure layer, Telslian's self - developed Hybrid Inference Intelligent Computing Platform (HICP) can achieve unified scheduling and management of multi - heterogeneous computing power, and can provide high - performance, low - latency computing resource support for cloud - based training and edge - side reasoning of embodied intelligence. Currently, the products at the AIoT infrastructure layer have completed compatible adaptation with more than a dozen domestic mainstream Xinchuang chips.

At the AIoT intelligent agent layer, the intelligent agents created by Telslian have four capabilities: human - like thinking, long - term memory, high - dimensional perception, and team collaboration. They form an efficient collaboration with Ubtech's swarm intelligence system, inject a "smart brain" into robots, enable robots to leap from "mechanical execution" to "intelligent collaboration", and significantly enhance their autonomous decision - making and collaborative operation capabilities in complex unstructured environments.

The scene of the strategic cooperation signing ceremony between Telslian and Ubtech

Zhou Jian, the founder, chairman of the board, and CEO of Ubtech Robotics, pointed out that for robots to evolve from single - execution tools to intelligent agents with human - machine co - integration capabilities, they must be deeply integrated with the understanding of environmental semantics and dynamic laws.

In this cooperation, Telslian's large spatial model and high - quality spatial data can provide sufficient training support for robots. The multi - modal interaction data generated by Ubtech's humanoid robots during operation can also continuously enrich the spatial data resources, further feed back and optimize the spatial reasoning ability of the model, promote the continuous evolution of the system, and jointly build a next - generation intelligent agent robot with a high degree of autonomy and interaction ability in the physical world.