FLEXOO and MIRMI join forces on Physical AI & robotics
2026-10-22
Strengthening collaboration at the intersection of robotics, AI, and sensing
FLEXOO and the Munich Institute of Robotics and Machine Intelligence (MIRMI) at the Technical University of Munich (TUM) have signed a Memorandum of Understanding (MoU), establishing a collaboration in the field of Physical AI.
The agreement builds on a shared interest in advancing intelligent systems that can better understand and interact with the physical world. By combining expertise in tactile sensing, robotics, machine intelligence, and real-world applications, the collaboration aims to explore new approaches for future Physical AI systems.
As AI increasingly moves beyond the digital domain, access to high-quality physical interaction data is becoming a key enabler for next-generation intelligent machines.
Connecting physical data with intelligent systems
Physical AI requires systems that can perceive and interpret what happens during physical interaction. Many critical signals emerge at the point of contact between objects, tools, robots, and their environments. Force, pressure, friction, deformation, and material behaviour often contain information that cannot be captured through conventional approaches alone.
FLEXOO's ultra-thin tactile sensing technology enables the capture of such data directly at the point of interaction, creating new opportunities for intelligent systems to better understand and respond to the physical world. The collaboration with MIRMI provides a framework for exploring how these physical data streams can be combined with advances in robotics and machine intelligence.
Advancing embodied intelligence and Physical AI
MIRMI is TUM's institute for robotics and machine intelligence, bringing together expertise across robotics, AI, healthcare, human-machine interaction, and industrial applications. Under the leadership of Prof. Dr. Lorenzo Masia, Executive Director at MIRMI and one of Europe's leading researchers in embodied intelligence and human-centered robotics, MIRMI is advancing how intelligent systems learn from physical interaction.
The institute combines expertise in robotic systems, wearable robotics, human-machine interaction, and advanced data capture technologies. This includes the development of specialized data capture equipment such as instrumented gloves, wearable systems, and robotic platforms designed to generate highly structured datasets of physical interaction.
These capabilities create an efficient pathway from sensor and hardware development to curated real-world datasets that can be used to train and validate future Physical AI systems. This is particularly relevant for embodied AI, where intelligent systems learn through interaction with the physical world and where access to large-scale robotic training environments is becoming increasingly important.
A prominent example is RoboGym, a joint initiative involving TUM, MIRMI, and NEURA Robotics. The project is designed to create one of the largest scientific training infrastructures for Physical AI and robotics worldwide, addressing one of the field's most important challenges: the generation of large-scale, high-quality real-world training datasets for embodied systems. By combining expertise in robotics, machine intelligence, data capture, and large-scale training environments, MIRMI provides a unique foundation for advancing Physical AI from research concepts towards real-world deployment.
The collaboration between FLEXOO and MIRMI brings together complementary capabilities in tactile sensing, physical data capture, robotics, and embodied intelligence to support the development of future Physical AI systems that can learn directly from physical interaction.
First joint activities for Physical AI
The first joint activities will focus on integrating FLEXOO's ultra-thin tactile sensing technology into selected RoboGym platforms, including humanoid robotic arms, multi-finger hands, and teleoperation systems. This will enable the capture of rich physical interaction data directly at the point of contact, including pressure, slip, and contact geometry - signals that are often inaccessible to vision-based systems alone.
Using these platforms, FLEXOO and MIRMI plan to generate large-scale datasets from real-world manipulation tasks such as grasping, insertion, and assembly operations. Particular attention will be given to challenging scenarios involving occlusions, deformable objects, and near-failure situations that are highly relevant for real-world robotic deployment. The resulting datasets will be used to train and evaluate FLEXOO's tactile-first Physical AI platform, and to benchmark tactile-first approaches against established vision-based methods.
A further objective is to explore cross-embodiment learning, enabling tactile knowledge acquired on one robotic platform to be transferred across different systems, including industrial grippers and multi-finger robotic hands.
By combining FLEXOO's expertise in tactile sensing and physical data capture with MIRMI's robotics infrastructure and RoboGym training environments, the collaboration aims to establish a long-term European foundation for tactile-first Physical AI research, development, and validation.
Exploring future robotic learning environments
The collaboration also creates opportunities to explore future approaches to robotic learning and embodied intelligence. Large-scale training environments such as RoboGym are expected to play an important role in the development of future Physical AI systems by enabling intelligent machines to learn from physical interaction at scale.
Combining these approaches with advanced tactile sensing and physical data generation opens exciting opportunities for the next generation of intelligent robotic systems.
Expanding opportunities for future innovation
Through this collaboration, both organizations aim to contribute to the development of intelligent systems that can learn from physical interaction and unlock new possibilities across robotics, automation, and other real-world applications.
MIRMI (Munich Institute of Robotics and Machine Intelligence) is TUM's interdisciplinary institute for robotics and machine intelligence, bringing together expertise across AI, robotics, healthcare, industrial applications, and human-machine interaction to advance the next generation of intelligent physical systems.