Why Tendon Drives Outperform Gear Trains in Dexterous Hands

Tendon-driven robot hands achieve higher dexterity-to-weight ratios than gear-driven alternatives by decoupling the mass of the actuator from the moving fingertip. Cable or wire elements — the tendons — are routed through a kinematic finger structure to transmit forces from remotely located actuators to distal joints. The actuator stays proximal; the fingertip stays light. This is the same biomechanical logic that governs the human hand, where forearm muscles drive finger movement through long tendons crossing the wrist.

The field is gaining renewed urgency in 2026 as humanoid robotics, minimally invasive surgical systems, and prosthetic applications all demand high dexterity in constrained form factors. Three major technical threads run through the patent record. First, underactuated tendon control — using fewer tendons than degrees of freedom — reduces mechanical complexity while force-based control eliminates slack. Second, tendon routing through slender surgical instrument shafts enables articulated wrist and jaw motion at scales inaccessible to gear trains. Third, cable-driven parallel platforms, first established by McDonnell Douglas in 1997 with an 8-tendon, 6-degree-of-freedom system, laid the kinematic foundations that later influenced hand and wrist designs.

Shape memory alloy wires represent a further departure from conventional motor-cable approaches: M Engineering Co.’s 2009 JP patent on a multi-finger movable robot hand uses SMA guide wires to drive bending and stretching of multi-joint fingers, with a mobile body in the wrist direction creating the towing operation for grasp and release. According to standards and research published by IEEE, cable-driven and tendon-actuated mechanisms continue to be among the most active research areas in robotic manipulation, reflecting the breadth of engineering approaches now converging on this architecture.

Three Decades of Innovation: From Cable Platforms to Biomechanical Control

The patent filing timeline in this dataset spans from 1997 to early 2026, and three distinct phases are clearly visible: a foundational phase establishing cable-driven kinematics, a development and specialisation phase advancing surgical and control theory, and a current acceleration phase embedding tendon mechanisms into humanoid and prosthetic systems.

The foundational phase (1997–2009) established cable-driven multi-DOF kinematics through McDonnell Douglas’s 8-tendon, 6-DOF suspension platform (1997, JP) and introduced shape memory alloy wire actuation for finger bending through M Engineering Co.’s multi-finger robot hand (2009, JP). The development and specialisation phase (2013–2021) advanced control theory — General Motors’ 2013 CN filing on underactuated tendon torque control — and extended tendon architectures into catheter-scale flexible endoscopy systems (EndMaster, 2020, JP) and cable-driven surgical wrists (Verb Surgical, 2021, JP). Medical Microinstruments began filing microsurgical tendon-actuated instrument patents from approximately 2021 onward.

The acceleration phase (2022–2026) shows tendon mechanisms becoming embedded components within sophisticated robotic surgical instruments and humanoid hand systems. Zhejiang Laboratory (a Chinese national research laboratory) filed two robot hand patents in 2025 covering multi-axis finger kinematics consistent with tendon routing architectures. Massachusetts Institute of Technology filed in 2024 on muscle-tendon state estimation for wearable robots, and Medical Microinstruments continued filing on tendon contact surfaces and link geometry. As noted in research tracked by WIPO, robotics and surgical instruments are among the fastest-growing patent technology areas globally, consistent with the acceleration observed in this dataset.

Four Technical Clusters Shaping the 2026 Landscape

The 14 patents in this dataset organise into four distinct technical clusters, each addressing a different engineering challenge in tendon-driven robotic systems. Understanding these clusters is essential for R&D teams deciding where to invest and where to file.

Cluster 1: Underactuated Tendon-Driven Finger Control

This approach uses fewer tendons than degrees of freedom, applying force control — not position control — to eliminate unconstrained slack. General Motors’ 2013 CN patent discloses a robot with tendon-driven fingers using n tendons for n DOF, with asymmetric joint radii enabling independent joint torque assignment. Saito Yusuke’s 2019 JP patent extends this with a passive coupling mechanism: drive link, tip link, and restriction link geometry provides automatic distal joint curl following proximal joint rotation, reducing the number of required actuators without sacrificing finger dexterity.

“Asymmetric joint radii enable independent torque commands at individual joints — a key enabler for precision grasp that eliminates the need for one actuator per joint.”

Cluster 2: Tendon-Actuated Surgical Instrument End Effectors

In this cluster, fine tendon wires are routed through slender shafts to actuate articulated wrist joints and jaw or gripper elements at the distal end of minimally invasive surgical instruments. Medical Microinstruments’ 2024 JP filing establishes unambiguous correlation between electric actuator motion and tendon displacement during surgical operation phases — applicable to multi-tendon instruments with 6 or more independent tendons. A companion 2023 JP filing engineers tendon contact surfaces on link structures to avoid bore-type routing, which generates higher friction and wear. CMR Surgical’s 2020 JP patent embeds multi-axis torque sensors in the intermediate segment of a four-axis wrist, enabling feedback-controlled tendon tension management.

Cluster 3: Flexible Tendon Robotic Systems for Endoscopy and Catheter-Scale Devices

Tendon-driven mechanisms applied to flexible, elongated shafts — endoscopes and catheters — solve a problem that rigid gear trains cannot: routing actuation through a flexible body without sacrificing controllability. EndMaster’s 2020 JP patent controls each robotic degree of freedom with a pair of actuators and a corresponding tendon pair, with programmable pre-tensioning that eliminates slack automatically. Verb Surgical’s 2021 JP cable-driven surgical wrist generates a third control command specifically to prevent cable slack — a direct parallel to EndMaster’s pre-tension approach, filed independently by a different assignee. This convergence from multiple independent organisations signals that slack management in flexible tendon systems is an active, unsolved engineering challenge.

Cluster 4: Muscle-Tendon Interface Control for Wearable and Prosthetic Robots

The most recent cluster applies biomechanical models of human muscle-tendon units as real-time control inputs for exoskeleton and prosthetic hands. MIT’s 2024 WO patent describes a muscle-tendon interface that measures physiological signals, with a torque set-point processor estimating muscle-tendon state corresponding to human motor intent. Closed-loop torque control then applies augmentation joint torques to wearable robot joints — enabling intent-driven, biomechanically coupled prosthetic control rather than surface EMG alone. Research published by institutions including NIH has documented the limitations of surface EMG as a prosthetic control signal, making MIT’s physiological tendon force sensing approach a significant advance in the field.

Geographic and Assignee Concentration: Who Holds the IP

Innovation in the tendon-driven robot hand dataset is not concentrated in a single dominant assignee but is distributed across medical device companies, academic research institutions, and defence and industrial primes. Japan (JP) accounts for the large majority of filing destinations across all retrieved results — reflecting Japan’s role as a major patent validation market rather than necessarily the origin of invention.

The core tendon-hand innovations originate from a diverse set of national entities. Medical Microinstruments S.p.A. (Italy/US) is the most active tendon-specific surgical instrument filer in this dataset, with at least 3 distinct tendon end-effector and microsurgery patents filed between 2021 and 2024. EndMaster Private Limited (Singapore) contributes the tendon-driven flexible endoscopic robotic arm with programmable pre-tensioning (2020, JP). General Motors Global Technology Operations (US/CN) established foundational underactuated tendon-finger torque control theory in 2013 under a NASA Space Act Agreement. Verb Surgical Inc. (US) filed on cable-driven surgical wrist control with slack prevention (2021, JP), and CMR Surgical Limited (UK) addressed torque sensing in tendon-actuated surgical wrists (2020, JP).

On the research institution side, Massachusetts Institute of Technology’s 2024 WO filing on muscle-tendon interface wearable robot control introduces a new IP moat for prosthetics based on physiological tendon force sensing. Zhejiang Laboratory (China) — a Chinese national research laboratory — filed two humanoid robot hand patents in 2025 covering multi-axis finger kinematics, signalling China’s national research investment in dexterous humanoid hand technology. Patent filing data tracked by the EPO confirms that China has become one of the most active jurisdictions in robotics patent filings globally, consistent with Zhejiang Laboratory’s international filing activity.

Emerging Directions and IP White Spaces

The most recent filings (2023–2026) in this dataset signal four forward-looking directions that R&D teams and IP strategists should prioritise. Each represents either an accelerating technical trend or an under-claimed area where novel filings can establish blocking positions.

Biomechanical signal-driven tendon hand control. MIT’s 2024 WO filing is the clearest indicator of a shift toward closed-loop torque control driven by real-time physiological tendon force measurements — not just kinematic tracking. Any prosthetic or exoskeletal hand company not engaging with biological tendon force sensing as a control input risks being locked out of the most physiologically natural control paradigm as MIT’s filing matures into national phase filings. The approach measures muscle-tendon state corresponding to human motor intent and applies augmentation joint torques accordingly.

Humanoid robot hand finger multi-axis kinematics. Zhejiang Laboratory’s two 2025 JP filings on robot hands describe a drive structure enabling simultaneous rotation about a palm-plane axis and a swing axis perpendicular to it — matching human metacarpophalangeal joint biomechanics. This two-DOF-per-finger architecture is directly compatible with tendon routing. IP strategists should monitor CN and JP filing families from this assignee for blocking positions in two-DOF-per-finger designs, as China’s national laboratory investment in humanoid hand kinematics is accelerating.

Tendon friction reduction through contact surface geometry. Medical Microinstruments’ 2023 filings explicitly engineer tendon contact surfaces on link structures to avoid bore-type routing, which generates higher friction and wear. This geometric optimisation approach — rather than material substitution — is an emerging sub-field for improving tendon efficiency in surgical-scale instruments. The specific claims around non-bore tendon contact surfaces suggest that link geometry for tendon routing is a patentable differentiation layer that most competitors have not yet claimed systematically, representing an IP white space.

Unambiguous multi-tendon actuation correlation for surgical instruments. The 2024 Medical Microinstruments JP filing addresses a long-standing challenge: establishing one-to-one mappings between electric actuator motions and tendon displacements in multi-tendon instruments with 6 or more independent tendons during live surgical use. This represents a move toward deterministic, verifiable tendon control suitable for regulatory approval — an engineering requirement that will become increasingly important as surgical robotic systems face scrutiny from regulatory bodies. Guidance on software and control validation for surgical robots is tracked by organisations including ISO under standards for medical electrical equipment and surgical robotics.

“Tendon contact surface geometry is an under-patented white space — Medical Microinstruments’ specific claims around non-bore contact surfaces suggest that most competitors have not yet claimed this differentiation layer systematically.”

For R&D teams targeting general-purpose robotic hands, the strategic implication is clear: surgical robotics is the commercialisation beachhead. The most mature and active tendon-hand patent families in this dataset are in minimally invasive and microsurgical applications. The friction management, slack control, and multi-tendon decoupling solutions developed in the surgical context are directly transferable to humanoid and prosthetic domains. Teams that study and build on this body of IP will have a significant head start over those developing tendon systems in isolation from the surgical robotics literature. Further context on the global robotics innovation landscape is available through PatSnap’s innovation intelligence resources and through the PatSnap R&D intelligence platform.