Key Takeaways
During the 2024 edition of NVIDIA’s annual GPU Technology Conference (GTC), robotics and applied AI dominated the conversation.
Nvidia’s biggest product announcement of the event – the Isaac Perceptor – is geared toward Autonomous Mobile Robots (AMR). Meanwhile, President and CEO Jensen Huang used his keynote address to speculate on the future of robotics.
Predicting that in the future “everything that moves will be robotic” Huang explained how NVIDIA is equipping robots with the tools to navigate complex spaces.
Observing that automation is generally “safer and more convenient,” he said self-driving cars will be “one of the largest” applications of robotics.
Nvidia’s latest product release is oriented toward industrial and manufacturing robots.
Currently, most industrial robots operate in specific structured zones, typically warehouse floors devoid of human workers and other potential hazards. But Nvidia’s Issac Perceptor envisages a future in which robots can work safely in unpredictable, changing environments.
Robots programmed with the new multi-sensor compatible software development kit can generate dynamic 3D models of their surroundings in real-time, helping them to avoid obstacles and respond to unexpected scenarios.
A similar technology – the Isaac Manipulator – is focused on enhancing robotic arms used in manufacturing, which Huang observed are the “most used robots today.”
Juang speculated that humanoid robots could perform a much wider variety of tasks than static arms or automated warehouse vehicles.
While Nvidia is busy developing the tools to build better self-driving cars and industrial robots, it is preparing for the future of robotics
Predicting that “the next generation of robotics will likely be humanoid,” Huang explained that more human-like robots have several advantages.
Teaching humanoid robots to perform complex tasks is easier because there is a wealth of movement data from actual humans they can be trained on, he said.
Robots that closely resemble the human form “will be much more useful in our world because we created the world to be something that we can interoperate in and work well in”, he added.
“We set up our workstations and Manufacturing and Logistics […] for humans. They were designed for people and so these human robotics will likely be much more productive to deploy.”
To help prepare for the future of humanoid robots, Nvidia has developed GROOT, a foundation model that learns from watching videos of human movement.
For the next stage of training, NVIDIA even created the “Isaac reinforcement learning gym, “which allows the humanoid robot to learn how to adapt to the physical world” Huang explained.