Immersiverse S.L

ImmersiVERSE is a Barcelona-based Physical AI and robotics technology company building the simulation infrastructure for intelligent machines. We develop physics-aware virtual environments where robots, autonomous vehicles, and AI systems can be trained, tested, evaluated, and prepared for deployment in the real world.

Our core technology combines high-fidelity digital twins, physics-based simulation, multimodal sensor simulation, synthetic data generation, reinforcement learning, imitation learning, and XR-based teleoperation. We recreate complex physical environments and simulate how robots and autonomous systems interact with the world, including their movement, perception, sensing, materials, environmental conditions, and physical interactions.

ImmersiVERSE develops simulation environments for training and validating Physical AI across navigation, perception, manipulation, grasping, autonomous mobility, inspection, and human-robot interaction. Our systems can generate realistic sensor data from LiDAR, radar, RGB cameras, depth cameras, stereo cameras, thermal sensors, IMU, GNSS, and other sensing modalities. This enables AI models to learn from large-scale synthetic experiences while reducing the cost and limitations associated with collecting and labelling real-world data.

A key part of our technology is physics-aware simulation. Instead of treating the environment as a purely visual representation, we model physical behaviour and environmental effects. This includes materials, friction, collisions, dynamics, weather, wind, rain, fog, snow, mist, visibility changes, and other challenging conditions. These environments allow autonomous systems to be exposed to scenarios that can be difficult, expensive, or unsafe to reproduce in the real world.

We are also building reinforcement learning and simulation-based training infrastructure for robots and autonomous agents. Our environments enable agents to learn through millions of simulated interactions, allowing navigation, decision-making, manipulation, and control policies to be developed and evaluated before deployment. We focus on scalable simulation-to-real workflows, domain randomisation, synthetic data, and realistic sensor modelling to help reduce the gap between virtual training and physical operation.

ImmersiVERSE is developing tools and platforms that connect simulation, AI, robotics, and real-world hardware. This includes custom robotics training environments, digital twins of robots and industrial spaces, multimodal sensor simulation, synthetic 2D and 3D ground-truth generation, physics-aware environmental systems, and interfaces for connecting simulation with robotics frameworks and physical platforms.

Another area of development is XR teleoperation and human-in-the-loop robotics. We use immersive interfaces, hand tracking, motion capture, and XR technologies to allow humans to interact with and control robotic systems remotely. This creates a bridge between human expertise and autonomous robotic systems and can support applications where fully autonomous operation is not yet practical or where human supervision is required.

Our technology is designed for applications across industrial automation, logistics and warehousing, autonomous mobility, robotics, inspection, healthcare, space, disaster response, hazardous environments, and other mission-critical applications. By moving more of the training, testing, and validation process into realistic simulation, we aim to make the development of Physical AI systems faster, more scalable, and more accessible.

ImmersiVERSE is ultimately building a simulation-first ecosystem for Physical AI: a place where robots and autonomous systems can experience realistic worlds, generate data, learn from interaction, be tested under challenging conditions, and transition from simulation to the physical world.