Nvidia robotics engineer Singh touts open source stack

Nvidia has outlined the work of Jaiveer Singh, the robotics software engineer leading development of Isaac ROS, the company's open source robotics software stack built on ROS 2.

Singh leads the team developing software for autonomous mobile robots, manipulation systems and humanoids using Nvidia hardware and software tools. The platform combines CUDA-based libraries and AI models with ROS 2, the open source Robot Operating System framework widely used in robotics development.

His account offers a look at how Nvidia is positioning itself in robotics as interest grows in physical AI and humanoid machines. Rather than focusing on robot demonstrations, he described the work as centred on infrastructure, including the computing boards inside machines, software that processes data from robot cameras and the engineering needed to move a system from prototype to repeatable deployment.

Isaac ROS began as an internal experiment during Singh's internship with the robotics team. After studying electrical engineering, computer science and business at the University of California, Berkeley, he joined Nvidia full time and continued work on the project that became the company's robotics software stack.

A key part of the approach is open source distribution. Isaac ROS was designed to let developers inspect, adapt and extend the software rather than depend on a closed system that might not meet their needs over time.

"My goal is to make sure everyone feels like they are a part of the robotics future," said Jaiveer Singh, Robotics Software Engineer, Nvidia.

Singh said the original question behind the project was whether releasing robotics software built around the Nvidia Jetson platform and Nvidia CUDA libraries as open source would create value for developers. He said the answer became clear once developers could use more of the processing power available in their graphics hardware.

"We wanted to see what would happen if we just released some software as open source that uses the NVIDIA Jetson platform and NVIDIA CUDA libraries for robotics. Would there be any value there?" said Singh.

"And the answer was, of course, yes, because developers always want to be able to unlock the full power of their GPUs," he said.

Modular design

Isaac ROS supports several categories of robotics development, including manipulation, mobility and humanoids. The software includes packages for perception, object detection, mapping, collision detection and motion planning, and it can run across workstations, DGX Spark systems and Jetson edge devices.

Singh said the software was designed as a modular system rather than a single fixed stack. That allows developers to combine Nvidia's software packages with existing ROS code created internally or drawn from the wider robotics community.

"Compared with the original Isaac SDK, Isaac ROS is completely modular," said Singh.

"We ship the software like a bunch of LEGO bricks - you get to assemble them however you want, and you can easily combine our packages with existing ROS code written by you or others in the global robotics community," he said.

Open source bet

The open source model has become central to Nvidia's case for Isaac ROS, particularly in a sector where product requirements and technical standards can shift quickly. Singh said developers and startups need confidence that the software they adopt will remain usable and adaptable several years into a project.

That issue is especially relevant in robotics, where companies often build products over long development cycles and must integrate software with sensors, actuators and safety systems. In that environment, the ability to inspect code, modify it and contribute fixes can matter as much as access to the original tools.

"The main reason open source is valuable is because it gives people confidence that they can build upon this stack at this very initial stage," said Singh.

"Because the entire landscape can shift so rapidly, developers need the confidence that this platform is still going to be there to modify and improve two or three years into the future," he said.

Nvidia is also adapting Isaac ROS for a market increasingly focused on humanoid robots and AI agents. Singh said his team has been making the software better suited to developers building those systems, which typically require an end-to-end software stack capable of handling perception, planning and control across different hardware environments.

He also pointed to Nvidia's earlier work in robotics as a factor that drew him to the company. In his view, Nvidia had invested in the field before today's surge of attention around physical AI.

"NVIDIA was here and working on this problem before anybody else thought it was important," said Singh.

"We already had a stake in the ground," he said.

For Singh, the broader case for open source in robotics goes beyond software licensing. He described it as a way to spread both confidence and responsibility across the industry, allowing one company's fixes and improvements to be used by others facing similar technical problems.

"When more people can build robots," said Singh, "the future gets here faster."