Simulation is crucial for the training, testing, and deployment of autonomous systems, but achieving real-world fidelity poses significant challenges. This requires precise modeling of the physical characteristics and behaviors of autonomous system sensors and their surrounding environments. Unveiled at NVIDIA GTC, the Omniverse Cloud API aims to address this challenge by providing large-scale, high-fidelity sensor simulation, accelerating the path to autonomy. These APIs aggregate a rich ecosystem of simulation tools, applications, and sensors.
These APIs meet the critical need for high-fidelity sensor simulation, allowing for the safe exploration of countless real-world scenarios that autonomous systems may encounter. Additionally, the Omniverse Cloud platform offers developers many powerful features such as OpenUSD for universal scene description, RTX, and cloud APIs supporting generative AI, fostering interoperability and physically accurate rendering for the next generation of tools.
Simulation is crucial for enhancing safety levels.
As demand for robots, autonomous vehicles, and other AI systems increases, developers are seeking methods to expedite their workflows. Sensor data boosts the perceptual capabilities of these systems, enabling them to understand their surroundings and make informed decisions in real-time.
Traditionally, developers have relied on real-world data for training, testing, and validation. However, this approach has limitations when it comes to covering rare scenarios or data that cannot be obtained in the real world. Sensor simulation provides a perfect way to effectively test countless "what-if" scenarios and various environmental conditions.
With the Omniverse Cloud API, developers can utilize high-fidelity sensor simulation to enhance their workflows, addressing the challenges of developing full-stack autonomy. This not only streamlines and improves the efficiency of the development process but also lowers the barrier to entry for companies developing autonomous machines.
Ecosystem Advantages
The Omniverse Cloud API creates a universal AI system development environment by bringing together simulators, validation and verification (V&V) tools, content, and sensor developers into a vast ecosystem. Developers and software vendors like CARLA, MathWorks, MITRE, Foretellix, and Voxel51 emphasize the broad appeal of these APIs across the entire autonomous vehicle industry.
CARLA, an open-source autonomous vehicle simulator used by over 100,000 developers, can enhance its existing workflow with high-fidelity sensor simulation using the Omniverse Cloud API.
MITRE, a nonprofit organization dedicated to improving technical security, is also establishing a digital testbed for the autonomous vehicle industry to validate autonomous solutions. This digital testbed will leverage the Omniverse API to provide core sensor simulation capabilities for its developers.
MathWorks and Foretellix provide essential simulation tools for designing, executing, monitoring, and debugging test scenarios. As demonstrated at the GTC conference, combining such simulation and testing automation tools with the API can build a robust testing environment for autonomous vehicle development.
Furthermore, by integrating the API with Voxel51's FiftyOne platform, developers can easily visualize and organize ground-truth data generated during simulation, simplifying and enhancing the efficiency of training and testing.
Leading industrial sensor solutions provider SICK AG is working to integrate these APIs into its sensor development process to reduce the number of physical prototypes, achieve rapid iteration of design modifications, and validate final performance. Ultimately, developers of autonomous systems can utilize these validated sensor models in their applications.
Developers can also access sensor models provided by various manufacturers, including laser radar manufacturers such as Velodyne Lidar, Innoviz Technologies, Luminar, MicroVision, Robosense, and Seyond; visual sensor suppliers like OMNIVISION, onsemi, and Sony Semiconductor Solutions Corporation; as well as radar suppliers Continental, FORVIA HELLA, and Arbe.
Moreover, AI/ML developers can leverage these APIs to generate large and diverse synthetic datasets, providing crucial data for training and validating perception models used in these autonomous systems.
Empowering Developers and Accelerating Innovation
By lowering the traditional barriers to high-fidelity sensor simulation, the NVIDIA Omniverse Cloud API empowers developers to tackle complex AI problems without the need for extensive infrastructure modifications. This democratization of advanced simulation tools is poised to accelerate innovation, helping developers quickly adapt to the latest technological advancements and integrate them into their testing and development workflows.
Apply for early access to the Omniverse Cloud API.
