Astribot Brings T1 to North America, Showcasing a Physical AI Stack Built on Design for AI

Astribot T1 Humanoid Robot

Astribot T1 at IROS 2026 Pittsburgh

Astribot at IROS 2026

At IROS 2026, Astribot presents its cable-driven T1 platform, Lumo AI models and embodied operating system, designed to turn intelligence into real-world action

PITTSBURGH, PA, UNITED STATES, September 29, 2026 /EINPresswire.com/ -- Astribot, a Shenzhen-rooted Physical AI company, is bringing its T1 humanoid robot to North America for the first time at IROS 2026, taking place September
27–October 1 at the David L. Lawrence Convention Center in Pittsburgh, Pennsylvania. The company will showcase its integrated approach to Physical AI spanning continuously improving AI models, an embodied operating system and a cable-driven robotic body. Rather than treating AI as a software layer added to an existing robot, Astribot follows a “Design for AI” philosophy, designing intelligence, the operating system and the physical body as an integrated system.

AI: Intelligence Designed to Act

Astribot places AI at the center of its Physical AI architecture. Its Lumo model family is designed to connect reasoning and action in physical environments, with the company developing models that can perceive, reason about and execute tasks through real-world robots. Lumo-1 introduced Astribot’s Reasoning-Action Foundation Model approach, while the company’s subsequent Lumo-2 work explores latent world-action modeling for more complex physical tasks. The goal is not simply to produce a stronger model in isolation, but to continuously translate advances in AI into new capabilities for physical
robots. Equipped with the Lumo-2 base model, T1 has demonstrated autonomous tidying of miscellaneous items into a backpack, combining decision-making with dexterous manipulation of deformable objects. Astribot is developing Lumo-2 to extend these capabilities to more complex household tasks; additional task capabilities depend on further training.

Embodied OS: Connecting AI to the Physical World

An AI model alone does not make a robot useful. Astribot is therefore developing an embodied operating system designed to connect AI capabilities with people, robots and physical environments while lowering the barriers to interaction, development and use. For developers, T1 provides SDK and API access to capabilities including joint control, Cartesian motion and whole-body coordination, as well as access to sensor data.

Astribot OS provides meta-packages and skill libraries that help developers orchestrate agentic behaviors and foundational functions. Its natural-language interface is designed to generate deployable robot applications from a single-line requirement. At the second Astribot OS Hackathon, which concluded September 21 in Beijing, 15 teams developed and demonstrated more than ten functional applications on T1 in 36 hours, according to the company.

Robotic Body: Designing the Physical Layer for AI

The third layer is the robotic body itself. T1 is built around Astribot’s cable-driven architecture, which is designed to support dexterous manipulation, compliant interaction and high-speed movement. The robot stands approximately 1.55 meters tall, weighs about 66 kilograms, has 23 degrees of freedom excluding end effectors, and supports up to 5 kilograms of payload per arm. T1 also supports automatic charging and quick battery replacement, while its modular design allows different end effectors, computing modules and sensors to be added for different tasks.

Astribot’s “Design for AI” approach reflects the company’s view that the physical body should be designed around what AI needs to perceive and do, rather than optimized independently from intelligence. In a cable-driven system, the company aims to combine dexterity, compliance and back drivability with the ability to expose richer physical interaction information to the intelligence stack. This makes the body more than an actuator platform: it becomes part of the system through which AI interacts with the physical world.

Learning by Doing

The three layers form a continuous development loop: AI generates capabilities, the embodied OS connects those capabilities to the robot and its users, and the physical body executes them in the real world. T1 does not learn autonomously in real time during deployment. Instead, Astribot collects multimodal data from robot operation and teleoperation, processes and curates that data, retrains models such as Lumo-2, and deploys updated skills through over-the-air updates or on-site installation and calibration. Customers can choose to collect data for dedicated model training or during autonomous task execution. For Astribot, this is the foundation of its approach to Physical AI: AI that grows by doing.

T1: Making Physical AI More Accessible

T1 is designed as a developer-accessible Physical AI platform for researchers, developers and real-world applications. With US pricing starting at $18,000, T1 combines modular hardware, developer interfaces, VR-based teleoperation and AI capabilities. US customers can order T1 now, with immediate delivery available.

About Astribot

Astribot is a Shenzhen-rooted Physical AI company developing AI models, an embodied operating system and robotic bodies as an integrated system under its “Design for AI” philosophy. Its T1 platform brings together cable-driven robotics and developer-accessible tools for physical AI research and applications.
Learn more at https://www.astribot.com/.

Lin Wan
Astribot
email us here