From Foundational Patents to AI-Native Tactile Sensing Architectures
Tactile sensing patent activity spans more than two decades, but the field has entered a distinctly new phase: the convergence of deformable optical sensors, spiking neural networks, and multi-modal AI fusion is reshaping what “touch intelligence” means for robots, surgeons, and XR users alike. The patent corpus analyzed here—active and pending filings through 2025–2026—reveals a technology moving from single-sensor characterization toward integrated sensing-feedback pipelines.
The early foundations (2003–2010) addressed the most fundamental problems: how to encode visual information near a surgical tool tip as haptic reaction force feedback (2003, JP), how to distribute tactile sensing across a robot body via self-organizing sensor networks (ATR, 2010, JP), and how to render continuous forces and torques in ungrounded haptic devices (AIST, 2008, JP). These filings established the conceptual vocabulary—sensing, encoding, rendering, distributing—that all subsequent work builds upon.
The mid-stage development period (2013–2020) saw maturation of capacitive touch sensing, haptic effect generation, and industrial robotic touch sensing. BL Autotech’s 2013 JP filing combined real sensor data with virtual models to present augmented reality haptic feedback for medical tools. Immersion Corporation refined surface-feature simulation through multiple friction-display filings in 2018–2019. Fanuc America Corporation filed on dynamic laser touch sensing for industrial workpiece registration in 2018.
The frontier filing period (2021–2025) is characterized by convergence of tactile sensing with machine learning and neuromorphic computing. National University of Singapore’s PCT filing in 2021 introduced spiking neural network (SNN) encoders for fusing visual and tactile modalities in real time, subsequently extended in active Japan filings through 2025. GelSight, Inc. filed two major Japan applications in April 2025 on touch sensing characterization and haptic intelligence integration.
Four Technical Clusters Defining the Tactile Sensing Patent Landscape
The tactile sensing patent corpus organizes into four distinct technical clusters, each addressing a different layer of the touch intelligence stack—from physical contact characterization to haptic rendering and industrial workpiece registration.
Cluster 1: Vision-Based Deformable Tactile Sensors
This is the most technically sophisticated and recently active cluster. A deformable elastomeric or gel-based surface is imaged from within; contact deformation is captured and processed to reconstruct contact geometry, force distribution, and surface texture with sub-millimeter resolution. GelSight, Inc. leads this cluster with its 2025 Japan filings: one covering a deformable transmissive layer coupled to an interface membrane illuminated internally (where a detector captures light patterns to determine surface orientation at each membrane contact point), and a second extending this architecture with secondary sensors and computing integration for haptic intelligence applications including remote telepresence. Mitsubishi Electric’s 2022 EP filing uses a camera to track displacement of marks on pins and ridges embedded in an elastomeric skin underside, with a pattern-matching algorithm identifying force distributions against a pre-learned library.
A sensing architecture in which a deformable elastomeric or gel-based surface is imaged from within by an internal camera or detector. Physical contact causes measurable deformation of the surface; the resulting light pattern or marker displacement is processed to reconstruct contact geometry, force distribution, and surface texture. GelSight’s architecture achieves sub-millimeter spatial resolution through this approach.
Cluster 2: Neuromorphic and AI-Fused Tactile-Visual Sensing
This cluster fuses tactile and visual sensing using event-driven architectures and spiking neural networks to enable real-time, power-efficient robotic perception. National University of Singapore’s classification system employs a first SNN encoder for visual event streams and a second SNN encoder for tactile event streams; a combination layer merges the modalities; a task SNN outputs a joint representation for object classification tasks such as container and weight identification. The PCT filing from 2021 has been extended through active Japan filings in 2023 and 2025, affirming continued SNN-based tactile-visual fusion as a strategic research direction. This approach eliminates the power and latency penalties of frame-based deep learning, making it viable for mobile and wearable robotic applications.
“The neuromorphic SNN tactile-visual fusion architecture from National University of Singapore eliminates the power and latency penalties of frame-based deep learning—making real-time touch intelligence viable for mobile and wearable robotic applications.”
Cluster 3: Industrial Robotic Touch Sensing and Workpiece Registration
Laser and wire-based touch sensing is used to locate workpieces in industrial automation, enabling robot programs to dynamically compensate for workpiece position offsets. Fanuc America Corporation’s 2018 Japan filing introduced a touch sensing plan that can switch between laser and wire sensing events mid-execution, with the workpiece offset updated and shared across multiple robot controllers. A 2022 Japan update reaffirms this multi-robot coordinate synchronization approach as an active asset. Fanuc Robotics America’s haptic teach pendant (2014, JP) closes the sensing-feedback loop at the human-robot interface by equipping the operator’s pendant with haptic feedback devices that alert upon haptic events.
Cluster 4: Haptic Rendering and Surface Texture Feedback
This cluster covers actuator-based systems that generate tactile sensations on touch surfaces. Immersion Corporation’s 2019 Japan filing describes a sensor that detects touch position on a surface, with a processor selecting a haptic effect—simulating virtual surface features—by varying the coefficient of friction via an actuator at the touch location. A 2020 Japan filing adds a macro fiber composite (MFC) element coupled to a touch surface that detects localized contact pressure and outputs a haptic effect at the same location, enabling position-specific tactile rendering. Rohm Co., Ltd.’s 2023 US filing acquires object-state information (position, direction, distance, speed, urgency) and presents this as pseudo-tactile stimuli on the user’s skin, encoding spatial and urgency dimensions through stimulus intensity and positioning.
Immersion Corporation’s friction-display haptic portfolio spans multiple Japan filing families from 2018 to 2020, covering the core mechanism of touch-location-dependent friction coefficient modulation. Any consumer device implementing variable-friction haptic surfaces faces a significant freedom-to-operate question against this portfolio.
Explore the full tactile sensing patent corpus and run freedom-to-operate analysis in PatSnap Eureka.
Analyse Patents with PatSnap Eureka →Where Tactile Sensing Technology Is Being Deployed
Tactile sensing patents are active across six distinct application domains, each with different maturity levels, IP density, and commercial urgency. Robotics and industrial automation represents the most patent-intensive domain, but medical devices, consumer electronics, XR, accessibility, and humanoid robot skin networks are all generating meaningful filing activity.
Robotics and Industrial Automation
Fanuc America’s laser and wire touch sensing system targets arc welding and assembly-line workpiece registration, with the 2022 Japan filing updating multi-robot coordinate synchronization. National University of Singapore’s SNN-based tactile sensor targets autonomous grasping and object classification. Mitsubishi Electric’s elastomeric sensor is positioned for robot end-effector contact characterization in manufacturing environments.
Medical Devices and Surgical Assistance
IntuiTap Medical’s 2022 Israel filing integrates a pressure sensor array with a needle guide carriage for real-time tissue pressure mapping during interventional procedures. The 2003 Japan smart tool patent explored converting visual information near a surgical tool tip into haptic reaction force feedback. BL Autotech’s 2013 Japan filing combined real sensor data with virtual models to present augmented reality haptic feedback for medical instruments. GelSight’s haptic intelligence system (2025, JP) explicitly envisions remote tactile examination of surfaces—with direct implications for telemedicine and remote surgical assistance.
Medical tactile sensing is described as underpenetrated relative to its technical readiness. IntuiTap Medical’s needle-guidance pressure-array system (2022, IL) and the earlier surgical haptic augmentation work (BL Autotech, 2013, JP) reveal a largely open IP landscape in procedural tactile guidance—identified as a high-value target for medical device companies and surgical robotics entrants.
Consumer Electronics and Human-Computer Interaction
Apple Inc.’s 2024 Japan filing links tactile output events to movement-threshold-based gesture recognition, creating nuanced confirmatory haptics on touch-sensitive surfaces. Immersion Corporation’s friction display portfolio provides surface texture simulation for consumer touchscreens. Sensel, Inc. filed on high-resolution tactile touch sensor arrays with physical overlays in 2022 (JP).
Accessibility, XR, and Humanoid Robot Skin Networks
The wearable tactile navigation system filed by Marc Hobein in 2013 (US) encodes GPS and compass data as skin-contact tactile nudges for eyes-free wayfinding. Rohm’s 2023 US filing encodes proximity object direction and urgency as spatiotemporal skin stimuli—directly applicable to obstacle warning for blind users. Disney Enterprises’ 2023 Japan filing links video saliency maps to haptic responses on touchscreens, enabling tactile engagement with video content. The 2025 Korea filing from Dalian Sichun Technology encodes XR palm-tracking trigger points for realistic hand-interaction simulation without physical controllers. ATR’s 2010 Japan skin sensor network—where individual tactile sensor nodes self-organize into a communication network to relay spatiotemporal data to a host computer—is regaining relevance with the rapid global rise of humanoid robot development, according to IEEE research on distributed robotic sensing.
Geographic and Assignee Concentration in Tactile Sensing Patents
Japan dominates filing volume in the tactile sensing patent dataset analyzed here. This reflects both the depth of Japan’s robotics and industrial automation sector—Fanuc, Mitsubishi Electric, ATR, Rohm, BL Autotech—and the strategic choice by US-origin companies including GelSight, Immersion Corporation, Apple, and Fanuc America to file extensively in Japan as a key robotics market.
International PCT filings signal broad protection intent: National University of Singapore’s foundational SNN tactile-visual paper was filed as a PCT application (WO, 2021), covering multiple potential national-phase entries. Europe (EP) appears for Mitsubishi Electric’s elastomeric tactile sensor (2022), targeting European manufacturing and robotics markets. South Korea shows active filings from Dalian Sichun Technology (XR touch simulation, 2025) and Korean research institutes on XR rehabilitation with tactile biofeedback. China is identified as an underrepresented jurisdiction relative to its known manufacturing and robotics scale—GelSight’s CN filing in 2024 and a CN robot motion planning filing suggest CN-market protection is beginning, but the domestic CN tactile sensing landscape likely contains substantial activity not reflected in this dataset.
According to WIPO‘s global patent filing data, robotics-related patent families have grown substantially in recent years, with Japan, the United States, and China representing the three largest filing jurisdictions for robotic sensing technologies. The concentration of tactile sensing filings in Japan within this dataset is consistent with broader trends in industrial robotics IP strategy documented by EPO patent landscape reports on emerging technologies.
| Assignee | Filing Focus | Jurisdictions |
|---|---|---|
| GelSight, Inc. | Vision-based deformable tactile sensors, haptic intelligence | JP, CN |
| Immersion Corporation | Haptic effect rendering, friction displays | JP |
| Fanuc America Corporation | Industrial robotic laser/wire touch sensing | JP |
| National University of Singapore | Neuromorphic tactile-visual SNN fusion | WO, JP |
| Apple Inc. | Consumer device tactile output and touch-sensitive interfaces | JP, US |
| Mitsubishi Electric | Elastomeric vision-based tactile sensors | EP |
| Rohm Co., Ltd. | Tactile stimulus presentation encoding | US |
| IntuiTap Medical | Medical pressure-array guidance | IL |
Innovation in tactile sensing is distributed across multiple players, with no single dominant assignee in core tactile sensing hardware. GelSight leads the vision-based deformable sensor segment; Immersion leads haptic rendering for consumer surfaces; Fanuc leads industrial robotic touch sensing; National University of Singapore leads neuromorphic tactile-visual integration.
Five Emerging Directions Shaping the Next Wave of Tactile Sensing
The frontier filings from 2021 through 2025 point clearly toward five convergent directions that will define tactile sensing technology over the next three to five years—moving from single-sensor characterization toward integrated, AI-native, multi-modal touch intelligence platforms.
1. Haptic Intelligence Platforms with Multi-Modal Sensor Fusion
GelSight’s 2025 Japan filings move beyond single-sensor tactile characterization toward integrated systems combining deformable optical sensors with secondary sensors—including cameras and LIDAR—for full surface-intelligence pipelines applicable to remote manipulation, quality inspection, and XR. This represents a shift from “what did the sensor feel?” to “what does the system understand about the surface?”
2. Neuromorphic Spiking Neural Network Tactile Processing
National University of Singapore’s progression from PCT (2021) through active Japan filings in 2023 and 2025 demonstrates a sustained commitment to event-driven SNN architectures that process tactile and visual data asynchronously. The 2025 Japan filing is the leading indicator of where neuromorphic robotic perception is heading. This architecture is consistent with broader research directions in neuromorphic computing documented by Nature on brain-inspired sensor fusion for autonomous systems.
3. XR-Integrated Tactile Interaction Without Physical Controllers
The 2025 Korea filing from Dalian Sichun Technology encodes palm-based XR interaction points for finger-trigger detection using multi-frame image tracking—signaling that XR hand-tracking and tactile simulation are converging toward a single interaction layer without physical controllers. This direction aligns with the broader trajectory of spatial computing platforms.
4. Medical Procedural Guidance via Tactile Pressure Arrays
IntuiTap Medical’s 2022 Israel filing integrates scanning pressure sensor arrays with needle guides and live imaging feedback—an early example of tactile sensing enabling real-time procedural navigation. This approach is likely to expand to catheter guidance, biopsy, and minimally invasive surgery more broadly, given the largely open IP landscape identified in this domain.
5. Tactile Content Encoding as a First-Class Media Channel
Disney’s 2023 Japan filing and Rohm’s 2023 US filing both indicate an emerging design space in which tactile feedback is treated as a first-class content channel—encoding video saliency, object proximity, or directional urgency as skin stimuli rather than merely as button-click confirmation. This direction has implications for media production, accessibility technology, and the design of next-generation human-machine interfaces.
Track neuromorphic tactile sensing filings and XR haptic patents as they enter national phase with PatSnap Eureka.
Explore Full Patent Data in PatSnap Eureka →Strategic Implications for IP and R&D Teams in Tactile Sensing
The tactile sensing patent landscape carries specific strategic implications depending on whether an organization is building robot grasping systems, consumer haptic interfaces, medical devices, or XR platforms. Four areas demand particular attention from IP strategists and R&D leaders.
GelSight’s patent moat in vision-based deformable sensing is deep and recently reinforced. The 2025 Japan filings extend a filing architecture that traces back through multiple generations of deformable transmissive layer claims. R&D teams building robot grasping systems should either pursue licensing, design around with fundamentally different physical sensing modalities such as piezoelectric arrays or capacitive skins, or build competing IP positions in adjacent claim spaces.
The NUS SNN tactile-visual fusion architecture is an emerging foundational framework for neuromorphic robotics. IP strategists should monitor the national-phase entry of the WO2021 and 2023 filings across US, EP, and CN jurisdictions where protection may still be pending or unopposed. Early engagement with this filing family—through opposition, licensing, or design-around—will be significantly less costly than post-grant challenges.
Medical tactile sensing represents the highest-value underpenetrated opportunity. The combination of technical readiness (demonstrated by IntuiTap Medical’s pressure-array guidance system and BL Autotech’s surgical haptic augmentation work) and a relatively open IP landscape makes procedural tactile guidance a high-priority target for medical device companies and surgical robotics entrants. The window for foundational claim filing in this domain may be narrowing as the technology matures.
China’s CN tactile sensing landscape is a significant intelligence gap. GelSight’s 2024 CN filing on haptic intelligence and a CN robot motion planning filing suggest CN-market protection is beginning, but the domestic CN tactile sensing landscape likely contains substantial activity not reflected in the dataset analyzed here. Any organization targeting Chinese manufacturing or consumer markets should commission a dedicated CN-jurisdiction analysis before making IP strategy decisions.
GelSight, Inc.’s patent position around deformable transmissive layer architectures for vision-based tactile sensing is described as a deep patent moat, recently reinforced by two major Japan applications filed in April 2025. R&D teams building robot grasping systems must either license, design around with different physical sensing modalities, or compete with fundamentally different architectures.