No longer a ‘Graphics’ company
NVIDIA has a long history of developing cutting-edge graphics processing units (GPUs) for the gaming industry, but in recent years, it has shifted its focus to other areas, including robotics. The company's latest offering is the NVIDIA Isaac software platform, which provides a comprehensive set of tools for developers to create and deploy intelligent robots.
On the hardware side, NVIDIA has focused on seamless camera devices connectivity, offering interfaces for 2D and 3D cameras and LiDAR devices connectivity. These products are sold under the Isaac Nova Orin hardware platform flag and target specifically the needs for mobile robot applications, both indoor and outdoor. In addition, NVIDIA made sure that the most common functions were offloaded to the GPU, freeing the CPU of inefficient image processing.
This full stack 2D/3D vision integration was a much-anticipated need in the robotics industry, and NVIDIA has delivered a solution that exceeds expectations. Other players in the AMR compute platform business, such as Intel and Beckhoff, have not prioritized this need to the same extent, which has given NVIDIA a significant advantage in the market. By providing a comprehensive solution that addresses the most pressing needs of the robotics industry, NVIDIA has positioned itself as the go-to platform for mobile robotics.
Now into the details !
The Isaac Software Platform
The NVIDIA Isaac software development kit (SDK) is a comprehensive set of tools and libraries designed to help developers create advanced autonomous robots. Some of the key benefits of the Isaac Software Platform include:
- High-performance robotics: The Isaac SDK is optimized for robotics applications, with support for real-time processing, high-speed communication, and low-latency control.
- Advanced perception: The SDK includes a range of algorithms and libraries for computer vision, depth sensing, and other perception tasks, enabling robots to understand and interact with their environment.
- Efficient simulation: The Isaac SDK includes Isaac Sim, a powerful simulation environment that allows developers to test and refine their robots in a virtual environment before deploying them in the real world.
- Modular architecture: The SDK is designed with a modular architecture, allowing developers to easily integrate their own algorithms and components and customize their robots to meet specific application requirements.
- Developer-friendly tools: The SDK includes a range of developer-friendly tools, such as a visual debugger and a code-generation tool, to help developers build and debug their robots quickly and efficiently.
An architecture for Mobile Robot Navigation
In addition to the strong tooling support, NVIDIA has also setup a modular architecture specific for autonomous robot control and navigation, and named that the Isaac Engine. Although the architecture is not fixed, a mobile robot wil typically be structured with the software modules as shown below.
We realize this is a very high level overview of the software features, but it sets you up to read another article we created that dives into a number of these software packages in our ‘Isaac ROS in under 5 minutes’ blog.
The Isaac Orin Hardware Platform
NVIDIA has also developed a range of hardware solutions specifically for robotics, including the Jetson family of embedded systems and the recently announced Nova Orin system-on-a-chip (SoC). These devices provide high-performance computing power in a compact form factor, making them ideal for use in mobile robots.
The Orin SoC is particularly noteworthy, as it represents a significant leap forward in terms of performance and power efficiency. It features a custom-designed ARM CPU, as well as multiple NVIDIA GPUs and dedicated hardware for deep learning and computer vision. This makes it an ideal platform for the development of advanced AMRs, such as autonomous delivery drones or warehouse robots.
In addition, NVIDIA claims that the stereo cameras and fisheye cameras are custom designed by NVIDIA in coordination with camera partners. All sensors are calibrated and time synchronized, and come with drivers for reliable data capture. Notable brands present in this list are StereoLabs, from the ZED 3D Stereo camera, Basler AG for industrial grade 2D cameras, and LIPS which offers 3D structured light, ToF and Stereo cameras.
The image below shows the interfaces that are ready-to-use for a typical AMR application:
- Emergency Stop (ESTOP)
- Stereo Camera and 2D camera over GMSL2 / DESER
- General Purpose Input/Output (GPIO) pins
- Ethernet for planar (2D) LIDAR
- 4 CAN controllers for various control functions of drive units, UltraSound Sensors (USS) and IO islands
- I2C lines for various IMU, Magnetometer and Barometer functions
Applications
Overall, NVIDIA's success in the field of mobile robotics can be attributed to its innovative technology, strategic partnerships with camera and LiDAR vendors, and commitment to supporting the development of autonomous systems. As the demand for AMRs continues to grow, NVIDIA is focusing on solving these applications:
- Logistics and warehousing: to navigate complex warehouses and distribution centers, pick and pack goods, and optimize supply chain operations.
- Manufacturing: to perform a range of manufacturing tasks, from assembly to inspection, using 2D/3D camera technology and Deep Learning.
- Agriculture: to automate a range of agricultural tasks, from seeding to harvesting, and help farmers optimize crop yields and reduce labor costs.
- Construction: to assist with construction tasks, such as site inspection and surveying, and improve safety and efficiency on job sites.
- Healthcare: to assist with patient care, such as transporting equipment and supplies, and enhance the overall quality of care.
Read more about the announcement of NVIDIA on their blog or Contact US to get you started on this platform !
