Library, driver, and API integrated into a single

NVIDIA announced the expansion of its Isaac platform for building robot applications, and is expected to accelerate the development and deployment of robots across various industries. NVIDIA founder and CEO Jensen Huang unveiled the Isaac software development kit at the GPU Technology Conference, which integrates libraries, drivers, APIs, and various other tools into a single package.

With the newly released SDK, manufacturers, researchers, startups, and developers can more conveniently apply AI to next-generation robots for sensing, navigation, and control, saving hundreds of hours. The SDK provides a framework for managing communication and transmitting data within the robot architecture, and enables tasks such as adding sensors, managing sensor data, and controlling actuators to be performed easily in real time.

In particular, the inclusion of Isaac Sim in this SDK, a simulation environment that allows for the development, testing, and training of autonomous robots in virtual situations, is drawing attention. When the algorithm trained in the simulation is deployed to NVIDIA Jetson for AI computing on edge devices, the robot enters actual operation.

NVIDIA unveiled its first-ever reference design based on the Isaac SDK and named it 'Carter,' emphasizing the convenience of being able to transport objects while avoiding them. Carter, a delivery robot, can be seen at the NVIDIA booth in the GTC exhibition hall this week.

While the potential of the robotics field is limitless, training and testing actual robots requires significant cost and time, and in some cases, carries inherent risks. NVIDIA’s Isaac Sim is an evolved version of the existing Isaac Lab, enabling developers to train and test autonomous machines developed with the Isaac SDK under detailed and realistic scenarios.

This is made possible by using development tools suitable for high-fidelity simulation and more advanced real-time rendering. Since Isaac Sim fits perfectly with the tools and frameworks provided by the Isaac SDK, the transmission and reception of data and algorithms to and from actual robots proceeds seamlessly. This provides the critical prerequisite for very rapid robot development, allowing engineering and testing tasks that typically take several months to be completed in minutes. Once the simulation is complete, the trained system (brain) can be transferred to the actual robot.

The NVIDIA Isaac Platform, with the Isaac SDK and Isaac Sim integrated into Jetson, is expected to play an ideal role in the development and deployment of robotic applications in manufacturing, logistics, transportation, agriculture, services, construction, inspection, and various other fields.