Solutions
Tactile data,
for robots and for people
commissure works with companies and research institutions in two areas: tactile sensing for robotics and physical AI, and human-centred experience design. We do more than supply our own products — we take part in the form that suits a partner’s needs and stage, from experience design support to joint research and joint development. By staying alongside from technical design and consulting through to implementation, we bring haptic experience into real use together.
Services
Research and field work,
turned into teaching material
Adopting Physical AI means handling the target task, contact conditions, the data collected, the evaluation method and the machine within one design. Drawing on research from the University of Tokyo and joint field trials, the programmes work through that decision.
-
Physical AI Training
Choose from four programmes — theory, PoC design, collaborative robots and humanoids — according to the work you are responsible for.
- Foundation: Physical AI theory / adoption design and PoC
- By machine type: collaborative robots / humanoids
Products
Building the sense of touch
Our products are not standalone devices but a single connected tactile stack. Sensors turn contact into data; a common format carries that data into a robot learning pipeline; and touch is fed back to the hands of a remote operator. By developing sensing, integration middleware, generative AI and feedback devices as one, we support contact-rich work that vision and language alone cannot reach.
Sensing
A wearable sensor that combines muscle activity (FMG) with vibration sensing to record, hands-free, where and how much force a person is applying. It carries expert demonstration data into robot imitation learning and skill transfer.
Authoring
Middleware that connects a wide range of tactile sensors to robot learning pipelines. It unifies the data formats and time synchronisation that differ from sensor to sensor, supporting the path from plugging a sensor in to starting imitation learning with touch.
A diffusion model that generates haptic signals from natural language. It can be used to augment measured data or to design haptic feedback automatically. Presented at EuroHaptics 2026.
A development kit for Unity and ROS 2 that speeds up the implementation of haptic feedback in XR and robotics.
Presenting
A wearable haptic device that returns the sensation of force to the hands in real time. Haptic teleoperation improves both the precision of an operation and the quality of the demonstration data used for imitation learning, and extends the experience in immersive content.
A multi-channel haptic display platform that drives many channels in sync. From R&D to training and exhibitions, it can be configured flexibly for large-scale, high-density haptic display.
A chair-shaped device that delivers touch to the whole body through the seat. It supports experiences felt with the entire body, in skill transfer and in immersive content.
Technology
Making the world of contact learnable
Robots have gained eyes (cameras) and a brain (AI models). What they still lack is a sense of touch. Automating contact-rich work — gripping, pressing, fitting things together, which is most of what industry does — requires tactile data. commissure is founded on an understanding of human embodiment: an accumulated body of academic research into how the body touches the world, senses force and assembles movement. Starting from that understanding, we are building the technical foundation to collect, align, measure and learn from tactile and force data, on the ground in Japan.
Capture Collect
Tactile and force sensors are fitted to robots and to people on site, and recorded in time sync with video and joint data. A human demonstration and a robot’s own experience become the same kind of data.
※ Patented technology held by commissure
Integrate Align and measure
A sensor-independent common data format puts data from different sensors on the same footing. We assess quantitatively — from sync accuracy, dropouts and calibration state — whether recorded data is good enough to train on.
Learn & Feedback Train and return to people
Tactile data is integrated into imitation learning pipelines so that contact-rich tasks can be learned. At the same time, high-fidelity haptic display returns that sensation to people.
※ Patented technology held by commissure
Company
Creating a future where the joy of being able flows onward
We are an infrastructure company for bodily experience, working towards a future where the joy of being able flows onward — by making bodily sensation, which has been hard to share, something that can be captured, reproduced and generated.
- Mission
- Making bodily sensation something that can be captured, reproduced and generated, so that sensory assets can be put to use across society
- Vision
- Creating a future where the joy of being able flows onward
News
-
Releasing Tactile Arena, a comprehensive catalogue of tactile sensors for robots -
Two full papers accepted at EuroHaptics 2026, presented in Siena, Italy -
Releasing Awesome UMI, a comprehensive catalogue of UMI devices for robot imitation learning data -
Interview: note article “A future of exchanging sensations — can touch convey experience?” published -
Our website has been redesigned -
Releasing a developer and designer toolkit for the HaptoScala® multi-channel haptic display system