Decart lays the foundation for physical AI systems with Oasis 3

2026-06-10 · Source: Dataconomy · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Robotics & Autonomous Systems, Emerging Technologies & Innovation · Depth: Advanced, medium

Summary

Decart, an AI startup, has released Oasis 3, the third generation of its world model series, designed to lay the foundation for "physical AI" systems like autonomous robots and vehicles. This release follows Decart's recent \$300 million funding round from investors including Nvidia and Toyota. Oasis 3 significantly advances world models by introducing real-time, action-conditioned video generation, enabling hyper-realistic simulations crucial for scaling robotic reinforcement learning. Accessible via a live API, it generates "endless" 3D worlds at 22 frames per second and 768px resolution, with less than 200 milliseconds latency. Key improvements include accurate physics simulation and multiview camera synchronization, providing three perfectly aligned camera angles for depth perception. This infrastructure supports training for diverse applications, from self-driving cars navigating complex scenarios to humanoid robots developing fine motor skills.

Key takeaway

For Robotics Engineers developing autonomous systems, Oasis 3 fundamentally changes how you approach training. You can now generate infinite, hyper-realistic, physics-based environments with real-time feedback and multiview camera synchronization, eliminating the need for costly real-world scenario recreation. This allows you to rapidly scale reinforcement learning for self-driving cars, industrial drones, or humanoid robots, accelerating deployment and reducing development cycles.

Key insights

Oasis 3 provides hyper-realistic, real-time simulation environments critical for training autonomous physical AI systems.

Principles

• Physical AI requires realistic, scalable simulation environments.
• Real-time interaction and accurate physics are crucial for robot learning.
• Multiview camera synchronization enhances depth perception in simulations.

Method

Oasis 3 uses real-time, action-conditioned video generation to create "endless" 3D worlds, integrating advancements from previous Oasis versions and Lucy video models for physics-based environments with low latency and multiview camera synchronization.

In practice

• Integrate Oasis 3 API for scalable robotic reinforcement learning.
• Generate diverse driving scenarios for autonomous vehicle training.
• Simulate fine motor skill development for humanoid robots.

Topics

• Physical AI
• World Models
• Oasis 3
• Robotic Reinforcement Learning
• Real-time Simulation
• Autonomous Vehicles
• Generative Video

Best for: Computer Vision Engineer, AI Scientist, Research Scientist, Robotics Engineer, Machine Learning Engineer, AI Engineer

Related on AIssential

• What is Physical AI? How Robots Learn & Adapt in Real Life — Physical AI integrates perception, reasoning, and action in the physical world, driven by advanced models and efficient compute.
• AnchorWorld: Embodied Egocentric World Simulation with View-based Evolution Customization — A framework enables embodied egocentric world simulation with human motion control and localized, text-driven scene evolution using hybrid-view training.
• [P] I open-sourced a synth framework for creating physics-simulated humanoids in Unity with MuJoCo -- train them with on-device RL and interact in VR — This system enables physics-based humanoid creation and on-device reinforcement learning for mixed reality.
• From Seeing to Simulating: Generative High-Fidelity Simulation with Digital Cousins for Generalizable Robot Learning and Evaluation — Digital Cousins generates diverse, high-fidelity simulations from real-world panoramas for robust robot learning.
• NVIDIA Research Advances Robotics From Simulation to the Real World — Simulation-to-real transfer is foundational for achieving generalizable, reliable embodied autonomy in robotics.
• Table tennis robot defeats some of world’s best players – why this has major implications for robotics — Ace robot demonstrates AI's ability to achieve superhuman performance in complex, real-world physical environments.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Dataconomy.