Nvidia Builds Isaac Platform for Intelligent Robots

The software platform will be powered by the Jetson Xavier computer, which is designed for robots, drones and other autonomous machines at the edge.

Nvidia Jetson

Nvidia is rolling out a platform designed to bring greater artificial intelligence to robots in a wide range of industries and which is only the latest step by the company to make its GPUs the central computing technology for machine learning and other modern workloads.

The chip maker introduced the platform, named Isaac, over the weekend at the Computex show in Taiwan, with officials saying it will be the foundation of future autonomous machines—such as industrial robots and drones—in such sectors as manufacturing, construction, agriculture and logistics. The platform is based on Nvidia’s Jetson Xavier (pictured), a computer designed specifically for robots, with six different kinds of processors, more than 9 billion transistors and a performance of more than 30 TOPS (trillion operations per second).

Jetson Xavier has more processing capabilities than a powerful workstation but uses a third of the energy of a lightbulb, according to Nvidia officials.

“AI is the most powerful technology force of our time,” Nvidia CEO Jensen Huang, who introduced Isaac at the show, said in a statement. “Its first phase will enable new levels of software automation that boost productivity in many industries. Next, AI, in combination with sensors and actuators, will be the brain of a new generation of autonomous machines.”

The rapid growth of AI and such components as machine learning, deep learning and neural networks is fueling a wave of innovation that promises to change the way operations are done in a wide array of industries. Intelligent robots and other autonomous machines—such as self-driving vehicles—will play key roles in those changes. According to a report by Loup Ventures, the industrial robot market will grow 11.8 percent a year through 2025, when it will hit $33.8 billion. Thirty-four percent of the industrial robots sold by then will be collaborative machines—which Loup referred to as “co-bots”—made to work safely alongside humans in such settings as factories and warehouses.

A growing number of tech vendors, including CA Technologies, are putting a greater focus on collaborative robotics.

Nvidia executives several years ago said the company would target AI, machine learning and similar modern workloads as key growth areas for its GPUs, which have the parallel computing capabilities many of these applications need. Most recently, the company late last month unveiled the HGX-2, a server platform for both high-performance computing (HPC) and AI workloads. Also in May, Nvidia rolled out the Isaac robot simulator to enable robot makers to design and test scenarios for autonomous machines.

Now the GPU maker has the Isaac software platform for intelligent robots based on Jetson Xavier. The Isaac platform includes a software development kit, which offers a collection of frameworks, APIs and libraries for accelerating the development of algorithms and software for robots. It’s tightly integrated with Isaac Sim, another element in the platform that gives developers a realistic virtual simulation environment for testing, training and iterations. Isaac Intelligent Machine Acceleration (IMX) is a collection of robotics algorithm software developed by Nvidia that address such aspects as perception, navigation, manipulation and control.

Jetson Xavier is an embedded module system designed for robots, drones and other autonomous systems at the network edge. It comes in operating modes of 10, 15 and 30 watts and delivers 20 times the performance and 10 times the power efficiency of its predecessor, the Jetson TX2, according to company officials. The six kinds of processors on the computer include a Volta Tensor Core GPU, an eight-core, 64-bit Arm-based CPU and two NVDLA deep-learning accelerators.

In addition, there is an image processor, a vision processor and a video processor. The goal is to enable dozens of algorithms to be processed simultaneously and in real time, which translates into robots that can take in data from its sensors and make the proper decisions based on that data, which is important for ensuring that the machine can work safely next to humans, officials said.

Early access to the Jetson Xavier developer kit—including the Isaac software—will start in August, with the kit priced at $1,299.