Humanoid NVIDIA Collaboration: Jetson Thor

Humanoid is expanding its collaboration with NVIDIA to bring more computing directly onboard its next-generation humanoid robots. Announced in London on October 21, 2026, the work focuses on a robotic edge runtime that connects perception, reasoning and control in real time, according to Humanoid.

The UK-based robotics company is integrating NVIDIA Jetson Thor, Isaac OS, Holoscan Sensor Bridge and Holoscan SDK. The goal is to process sensor data and run increasingly sophisticated models onboard as its robots move toward more complex real-world applications.

NVIDIA Jetson Thor: onboard computing

Humanoid says the move to NVIDIA Jetson Thor has allowed it to consolidate two onboard computers into a single system while maintaining computing capacity. The company also sees potential for additional computing headroom in the future.

Jetson Thor's unified CPU and GPU memory is another part of the architecture. Humanoid says it reduces unnecessary data movement and latency while making it simpler for different workloads to share computing resources.

How NVIDIA Holoscan connects sensors and control

NVIDIA Holoscan Sensor Bridge is intended to shorten the path between robot sensors and onboard models. Humanoid says camera data can be placed directly into GPU-accessible memory, reducing copies and CPU workload across perception, inference and control.

Humanoid is integrating Holoscan SDK alongside NVIDIA Holoscan Sensor Bridge as part of the computing foundation. Together with the other NVIDIA technologies, the work is intended to coordinate sensor processing and onboard models in real time.

NVIDIA Isaac: from simulation to real-world operation

This collaboration builds on Humanoid's existing use of NVIDIA Isaac Lab and Isaac Sim to train and simulate models before deploying them on physical robots. The company describes a connected workflow from simulation through to real-world operation.

For the next-generation architecture, NVIDIA Isaac OS provides an open, deterministic, GPU-native runtime for coordinating robotics workloads. Humanoid says the broader work can reduce hardware and software complexity, making its platform easier to develop, test and scale.

NVIDIA collaboration: real-world applications

Humanoid's focus is not computing performance alone. Bringing sensing, reasoning and control together onboard is central to the NVIDIA collaboration as the company prepares its robots for increasingly complex real-world applications.

Founder and CEO Artem Sokolov describes NVIDIA as a technology partner from the early stages of Humanoid's development. His emphasis is on the computing power and architecture needed to bring sensing and control together in real time. The capabilities described here are Humanoid's account of its platform development, rather than independent performance measurements.

NVIDIA technologies

Which NVIDIA technologies is Humanoid integrating? Jetson Thor, Isaac OS, Holoscan Sensor Bridge and Holoscan SDK form the announced computing foundation.

How is Humanoid using NVIDIA Jetson Thor? The company says it has been able to consolidate two onboard computers into a single system and maintain computing capacity.

Does Humanoid already use NVIDIA simulation tools? Yes. The company uses Isaac Lab and Isaac Sim for training and simulation before deploying models on physical robots.

Source: Humanoid expands collaboration with NVIDIA to power its next generation of humanoid robots.

Image: Humanoid.