In a significant move that could reshape the entire robotics industry going forward, Taiwanese contract manufacturer Hon Hai Precision, better known as Foxconn, has agreed a strategic partnership with the world’s largest computer chip maker, Nvidia, to develop advanced humanoid robots.
This is according to a report on a Taiwanese news website, which appears to have got the information from Foxconn’s head honcho, the Hon Hai chairman himself, Young Liu.
This collaboration aims to combine Foxconn’s extensive manufacturing expertise and Nvidia’s cutting-edge artificial intelligence (AI) technologies to create robots capable of performing complex tasks across various industries.
Hai. How are you?
Foxconn, headquartered in Taiwan, is the world’s largest contract electronics manufacturer, renowned for assembling Apple’s iPhones and other devices.
Beyond consumer electronics, Foxconn has been a pioneer in automating its production lines, deploying a vast number of robots – often referred to as “Foxbots“ – to enhance efficiency and precision in manufacturing processes.
This extensive experience in robotics within its operations may mean that Foxconn is in a strong position to go on to become a formidable player in the development of humanoid robots.
However, this is not a guarantee, as neither company – Foxconn nor Nvidia – has a Steve Jobs at the helm and may end up designing and releasing a plug-ugly robot nobody wants.
Thunder and lightning
Central to the partnership between the two colossal companies is Nvidia’s upcoming release of the Jetson Thor computing system, scheduled for launch in the first half of 2025.
Jetson Thor is a next-generation compact computer designed specifically for humanoid robots, built on Nvidia’s Blackwell architecture, which is basically an AI superchip that aims to advance generative AI and accelerated computing.
Blackwell, which has 208 billion transistors, features an integrated functional safety processor, a high-performance CPU cluster, and 100Gb of Ethernet bandwidth, enabling the execution of complex AI models necessary for advanced robotic functions.
Such a powerful computing platform would allow humanoid robots to process vast amounts of data in real-time, facilitating sophisticated behaviours such as autonomous navigation, object manipulation, and human-robot interaction.
By integrating Jetson Thor into their designs, developers can create robots that are more adaptable and capable of performing a wide range of tasks in dynamic environments.
I wanna data model
The development of humanoid robots is further accelerated by the availability of large-scale datasets and advanced computing models. Notable systems recently reported on by RoboticsAndAutomationNews.com include:
AgiBot World: AgiBot has launched what it describes as “by far the largest humanoid manipulation dataset” in the world. This comprehensive ecosystem includes over 1 million trajectories from 100 robots across more than 100 real-world scenarios, designed to advance multi-purpose robotic systems.
ERA-42 by Robot Era: Robot Era has unveiled end large computing model for robotics. Combined with its self-developed five-finger dexterous hand, ERA-42 enables robots to perform over 100 intricate tasks, showcasing adaptability and precision in various operations.
Apptronik and Google DeepMind collaboration: Apptronik has partnered with Google DeepMind to integrate advanced AI with cutting-edge hardware, aiming to develop intelligent, versatile, and safe humanoid robots capable of operating in dynamic environments.
A future full of humanoids
The advanced hardware, comprehensive datasets, and sophisticated AI models is propelling the field of humanoid robotics toward unprecedented capabilities.
Humanoid robots equipped with powerful computing systems like Jetson Thor and trained on extensive datasets are expected to perform complex tasks across various sectors, including manufacturing, healthcare, and services.
This evolution signifies a shift toward embodied AI, where intelligence is integrated within a physical form, enabling robots to interact with and adapt to the real world in human-like ways.
Embodied AI involves the integration of perception, cognition, and action within a physical entity, allowing for more natural and efficient interactions with the environment.
The datasets and models being developed for humanoid robots are instrumental in advancing embodied AI as they provide the necessary information for robots to learn and perform tasks in a manner akin to human learning and behaviour.
The partnership between Foxconn and Nvidia exemplifies the collaborative efforts driving this transformation, combining manufacturing prowess with AI innovation to lead the next wave of humanoid robotics.
