KiQ Robotics Begins Physical AI R&D Combining Its Flexible Finger Robot Hand Tool

KiQ Robotics Co., Ltd., based in Kitakyushu, Fukuoka Prefecture and led by CEO Takashi Takimoto, has begun research and development in the field of Physical AI. The company will build on the expertise it has gained through its robot hand tool, the “lattice-structure flexible finger,” including multi-variety gripping and position-offset absorption. Using the portable data collection concept of the public research project Universal Manipulation Interface (UMI) as one foundation, KiQ Robotics aims to combine stable real-world contact enabled by its flexible fingers with staged real-machine evaluation, creating a Physical AI platform that can be easily verified in industrial settings. Interest has been growing in Physical AI, which performs tasks in the real world through robots and machines. For robots to handle objects on-site, recognition and motion planning are not enough; the final contact point, where the robot touches, grasps and moves the object, must also be stable. Industrial sites face challenges such as high-mix low-volume production, positional misalignment and variation in workpiece shape. Even if AI recognizes an object correctly, slipping or gripping failure at the final contact point prevents practical deployment. KiQ Robotics will address this “final contact” challenge by combining its flexible finger physical contact interface with a software platform for learning, evaluating and improving robot tasks. The flexible finger is a robot hand tool whose soft lattice-structure fingertips deform according to the shape of the workpiece and support objects through surface contact. It is already used in manufacturing sites for transporting diverse workpieces, reducing changeover work and lowering workpiece damage. In Physical AI, the key role of the flexible finger is to physically absorb positional shifts and shape differences, compensating for part of the contact precision required from AI control. KiQ Robotics positions the flexible finger as a real-world contact interface for the Physical AI era, helping robots not only “see, think and move,” but also stably “touch and grasp” real objects. The company’s development will use UMI as one conceptual foundation. UMI is a framework for collecting human work data without using the robot body itself and transferring it to real machines. KiQ Robotics will combine the contact characteristics of its flexible fingers, data quality management and staged real-machine evaluation. Its focus areas are stabilizing real-world contact, portable data collection and robot task learning, and data quality with staged real-machine evaluation. By using shape conformity and surface contact, the company aims to create conditions in which AI control is less likely to fail at the final contact stage. It will also collect demonstration data in environments close to actual worksites without occupying robot hardware for long periods, then advance task-specific robot learning and real-machine evaluation. Evaluation will proceed in stages such as stopped, non-contact, near-contact and contact, together with data quality checks. As a research use case, KiQ Robotics’ flexible finger is being used in artificial intelligence research at Professor Tamukoh’s laboratory at Kyushu Institute of Technology. The laboratory is collecting motion data with domestic service robots and values the flexible finger’s ability to grasp objects stably through surface contact, even when fingertip alignment is not exact. This serves as one example of the flexible finger’s role in Physical AI as the final physical contact point that supports AI action in the real world after recognition and decision-making. In the short term, KiQ Robotics will verify on real machines how the flexible finger absorbs positional and posture deviations during contact, and accumulate knowledge about the conditions under which tasks succeed when combined with AI control. The company will conduct PoCs for demonstration data collection, learning and real-machine evaluation, focusing on high-mix workpiece transport, gripping tasks involving positional offset and processes where workpiece damage must be avoided. In the mid to long term, KiQ Robotics will build a platform combining field data, evaluation gates and safety checks so that public robot foundation models and new imitation learning methods can be compared and used. The flexible finger is a robot hand tool developed and sold by KiQ Robotics. Its soft lattice-structure fingertips deform according to workpiece shape and support stable gripping through surface contact. Its features include reducing changeover work for diverse workpieces, stabilizing gripping by absorbing shape differences and positional offsets, reducing workpiece damage caused by direct contact with hard chuck jaws, and supporting both standard products and custom designs. As of April 2026, the product has been deployed in more than 130 cases, adopted by over 50 companies and achieved a repeat rate of about 30%. Going forward, KiQ Robotics will expand the flexible finger not only for gripping challenges in manufacturing, but also as a contact interface in the Physical AI field. KiQ Robotics is accepting inquiries regarding joint Physical AI demonstrations using the flexible finger, gripping tests, demonstration data collection, AI control validation and real-machine evaluation. The company designs verification themes according to on-site challenges. KiQ Robotics is a robotics startup based in Kitakyushu, Fukuoka Prefecture, focused on developing and selling the “flexible finger” robot hand tool, contracted development of dedicated equipment, and R&D in robot control and AI-related technologies.