New Delhi, India – – Indian AI startup AuraML has launched AuraSim, a multimodal world simulation model designed to accelerate the development of robotics systems. Built on NVIDIA’s Omniverse and Cosmos infrastructure, AuraSim aims to reduce the time and cost associated with training autonomous robots by creating high-fidelity digital twins before physical deployment. The launch took place alongside the India–AI Impact Summit 2026, under the auspices of the Government of India’s India AI Mission.
Bridging the Gap Between Virtual and Physical Robotics
AuraSim distinguishes itself through its ability to generate physics-ready, three-dimensional simulation environments from a variety of input types. Unlike traditional robotics simulation platforms that often require extensive manual modeling, AuraSim can ingest data from text descriptions, 2D floorplans, video walkthroughs, depth maps, and point clouds. This multimodal approach allows developers to rapidly prototype and test robotic behaviors in diverse and realistic scenarios.
The core benefit lies in the creation of these digital twins. By accurately replicating real-world environments within the simulation, developers can validate robotic models, identify potential failure points, and refine algorithms without the expense and risk associated with physical hardware. This is particularly crucial for complex robotic systems operating in dynamic and unpredictable environments.
Simultaneous Simulation for Complex Robotic Interactions
AuraSim isn’t limited to simulating a single type of robot. The platform supports the simultaneous operation of humanoid robots, autonomous mobile robots (AMRs), and stationary robotic arms within a shared virtual environment. This capability is significant because many real-world robotic applications involve collaboration between different types of robots. For example, a warehouse might utilize AMRs for transportation, stationary robotic arms for picking and packing, and potentially humanoid robots for more complex tasks requiring dexterity and adaptability.
AuraML’s agentic automation pipeline allows for the automated execution of thousands of scenario variations. Developers can modify text prompts to rapidly stress-test edge cases and improve the robustness of robotics foundation models. This automated testing process accelerates training cycles and enhances the safety and scalability of robotic deployments.
NVIDIA Partnership and the Global Physical AI &. Robotics Cohort
The development of AuraSim was closely aligned with NVIDIA, leveraging the capabilities of Omniverse and Cosmos. Tobias Halloran, Director of EMEAI Start-ups and Venture Capital at NVIDIA, noted the rapid growth of India’s AI startup ecosystem and NVIDIA’s continued support through initiatives like NVIDIA Inception. This collaboration provides AuraML with access to accelerated computing and scalable AI infrastructure, essential for handling the computationally intensive demands of high-fidelity simulation.
Alongside the platform launch, AuraML announced the Global Physical AI & Robotics Cohort, an eight-week mentorship-driven training program. Developed in collaboration with NVIDIA Inception, the cohort is designed to nurture the next generation of talent in physical AI and generative robotics systems. The program will offer hands-on experience with simulation-driven robotics development, targeting robotics engineers, AI researchers, and developers.
Positioning India at the Forefront of Robotics Innovation
Ayush Sharma, co-founder of AuraML, emphasized the company’s ambition to position India as a leader in physical AI, robotics simulation, and digital twin development. “By launching the first multimodal world simulation model from India, powered by NVIDIA’s world-class simulation technology, we are providing a global toolset to accelerate the development of advanced robotics foundation models everywhere,” Sharma stated.
AuraML’s core focus is on building foundational world models for next-generation robotics, utilizing generative AI and physics-based simulation. The integration of multimodal inputs, digital twin technology, and scalable infrastructure represents a significant step towards advancing robotics innovation, not only within India but on a global scale. As the Indian AI ecosystem continues to mature, platforms like AuraSim are poised to play a critical role in shaping the future of autonomous systems and industrial automation.
The platform’s ability to translate diverse data formats into usable simulation environments addresses a key bottleneck in robotics development – the creation of realistic and comprehensive training data. By lowering the barriers to entry for simulation, AuraML aims to empower a wider range of developers and researchers to contribute to the advancement of robotics technology.
