NVIDIA OSMO is a cloud-native orchestration platform designed to scale complex, multi-stage, and multi-container robotics workloads across on-premises, private, and public clouds. OSMO enables developers to orchestrate, visualize, and manage a range of tasks, including generating synthetic data, training models, conducting reinforcement learning, and implementing software-in-loop systems for humanoids, autonomous mobile robots (AMRs), and industrial manipulators.

Features

• Developer-Friendly Workflow Specification: Easily share accelerated computing clusters for multi-stage workflows across various compute nodes—all without needing Kubernetes expertise. Integrate OSMO into existing CI/CD pipelines to dynamically schedule tasks for regression testing, benchmarking, and model validation.
• Location-Agnostic Deployment: Deploy and orchestrate multi-stage workloads on Kubernetes clusters, supporting x86, Arm, and NVIDIA GPUs. It’s deployable on premises, as well as in AWS, Azure, Google Cloud, and NVIDIA Omniverse™ Cloud.
• Data Lineage and Asset Traceability: Easily set up YAML-based workflow tasks, from data generation to reinforcement learning, training, and model validation in simulation. Data traceability is available for auditing deployed models and maintaining data lineage for safety.
• Secured Services: Secure your solution with open standards, including OIDC support for authentication, account management, container registries, data storage, and secrets management.

Benefits

• Efficient Development: Reduces the complexity of managing multi-stage robotics workflows, enabling faster development and deployment.
• Enhanced Flexibility: Supports a wide range of deployment environments, making it adaptable to various infrastructure setups.
• Improved Traceability: Ensures data lineage and asset traceability, enhancing the safety and reliability of deployed models.
• Robust Security: Provides secured services with open standards, ensuring the protection of data and resources.