OSMO is a workflow orchestration platform that provides easy and efficient access to various types of compute and data storage solutions for robotics developers.
Welcome to NVIDIA OSMO
OSMO is an open framework for end-to-end robotics development scaled across heterogenous compute nodes.
OSMO enables robotics developers to scale the AI development seamlessly from PC or workstations to large sized compute clusters in the cloud
What's new
- Available Now as Open-Source on Github
- Run OSMO locally before scaling to the cloud
Key benefits
- Unify simultaneous execution across diverse compute nodes specialized for AI model training, simulation or physical AI runtime.
- Simple zero-code YAML based workflow configuration scalable with custom build systems or AI agents
- Connect any Kubernetes compute backend
- Integrate with to your existing services with open standards - identity provider, container registries and storage
- Deploy in air-gapped environments
Try OSMO
Deploy OSMO on brev.
Deploy on Cloud
Follow these steps for cloud deployment
-
Create required resources in the cloud for hosting OSMO using Terraform
-
Deploy and configure OSMO using our Deployment Guide
Documentation
If you are an AI / ML /robotics developer, follow our User Guide to run OSMO workflows
🤝 Support
For support and questions:
- Review our documentation at the links above
- Create issues on Github for support
Contribute to OSMO
To develop features in OSMO, you will need one of the below supported devices to build and run the service
License:
By pulling and using NVIDIA OSMO artifacts, you accept the terms and conditions of the License.
Suggested readings:
