Smart Leg Sleeve Technology

• Smart leg sleeves are lower-limb wearable systems that integrate sensors and actuators to estimate movement, monitor physiology, and assist joint function.
• They employ multimodal sensing and programmable mechanical interactions—using IMUs, sEMG, and strain sensors—to achieve precise motion estimation, sometimes with errors around 7.21°.
• They span applications from rehabilitation to mobility assistance and human-exoskeleton interaction, utilizing soft actuators and semi-rigid designs for effective knee support.

A smart leg sleeve is a lower-limb wearable system in sleeve form that integrates sensing, actuation, or programmable mechanical interaction to estimate movement, monitor physiological state, guide motor behavior, or assist joint function. Recent work spans intelligent knee sleeves for camera-free 3D lower-body motion estimation, multimodal sensing platforms for human-exoskeleton interaction, soft robotic knee-extension exosuits, semi-rigid knee assistive mechanisms, and soft sleeve actuators that conform to the limb while transmitting forces and moments efficiently (Zhang et al., 2023, Tang et al., 16 Aug 2025, Liu et al., 2024, Zhang et al., 14 Mar 2025, Abboodi, 8 Nov 2025).

1. Scope and system classes

The literature does not define a single canonical architecture for a smart leg sleeve. Instead, the term covers multiple device classes united by a sleeve-like wearable interface around the lower limb. Some systems prioritize sensing, such as the Texavie MarsWear Intelligent Knee Sleeves and the layered smart leg sleeve for physiologically informed human-exoskeleton interaction. Others prioritize assistance, such as the soft robotic exosuit for knee extension and the soft sleeve actuation architecture. A third group emphasizes hybrid mechanical interaction, including semi-rigid tendon-guiding devices and passive orthotic mechanisms (Zhang et al., 2023, Tang et al., 16 Aug 2025, Liu et al., 2024, Zhang et al., 14 Mar 2025, Lemire et al., 2020).

System class	Representative work	Reported function
Multimodal sensing sleeve	Intelligent Knee Sleeves (Zhang et al., 2023)	3D lower-body motion estimation
Layered physiological sleeve	Smart leg sleeve (Tang et al., 16 Aug 2025)	Exoskeleton interaction, effort optimization, injury risk detection
Soft assistive sleeve/exosuit	Hyper-bending exosuit (Liu et al., 2024)	Knee extension assistance during sit-to-stand
Soft actuation sleeve	Soft sleeve actuators (Abboodi, 8 Nov 2025)	Linear, bending, twisting, and omnidirectional motion
Semi-rigid knee sleeve device	Exo-Muscle (Zhang et al., 14 Mar 2025)	Deterministic knee load compensation
Passive orthotic add-on	Mechanical orthosis mechanism (Lemire et al., 2020)	Facilitate leg extension and knee locking

This typological diversity is significant because it locates the smart leg sleeve at the intersection of smart textiles, wearable robotics, rehabilitation engineering, and lower-limb biomechanics. A plausible implication is that “smartness” in this domain is not restricted to embedded electronics; it also includes mechanically programmed deformation, controllable force routing, and anatomically informed interfaces.

2. Sensing architectures and multimodal perception

The sensing-centered branch of smart leg sleeve research is organized around multimodality. The Intelligent Knee Sleeves use synchronized wearable and camera-based motion-capture recordings to estimate 3D lower-body motion from sleeve data alone. Each timestamp contains 32 features, including time, 14 pressure-sensor pins, two IMUs, and a relative quaternion, while labels are represented as lower-body joint quaternions derived from MoCap. Using fused IMU and pressure data, the system reports an average error of 7.21 degrees across all eight lower body joints, with better performance than IMU-only or pressure-only settings and with robustness to visual occlusion and field-of-view limitations (Zhang et al., 2023).

The layered smart leg sleeve extends this logic by mapping sensing modalities onto anatomical layers. IMUs target skeletal kinematics, textile-based sEMG electrodes target muscular activation, and ultrasensitive textile strain sensors target cutaneous deformation. The system is stitched into a commercial compression sleeve, includes 6 sEMG, 4 strain, and 4 IMU channels, and interfaces with an on-body Jetson Nano for real-time decoding. Its reported demonstrations are accurate ankle joint moment estimation with RMSE $= 0.13\ \mathrm{Nm/kg}$, real-time classification of metabolic trends with accuracy $= 97.1\%$, and injury risk detection within 100 ms with recall $= 0.96$, all validated on unseen users with a leave-one-subject-out protocol (Tang et al., 16 Aug 2025).

Consistent placement is a separate sensing problem addressed by sleeve design itself. In the inexpensive sEMG sleeve literature, 32 embedded monopolar electrodes are mounted in neoprene, with embedded grommets aligned to personalized skin markings such as moles, freckles, or scars. Reported performance includes a signal-to-noise ratio of $14.03 \pm 4.43$, donning time of $10.30 \pm 3.35$ seconds, and placement precision of $7.32 \pm 0.26$ mm. That work explicitly notes adaptability to lower-limb or “Smart Leg Sleeve” applications, suggesting that reproducible spatial registration of biosignals is a foundational requirement for lower-limb sleeve intelligence as well (George et al., 2020).

3. Actuation, force transmission, and assistive mechanics

Actuated smart leg sleeves differ primarily in how they generate and transmit forces to the limb. The soft robotic exosuit for knee extension centers on hyper-bending pneumatic actuators composed of a multi-material textile sleeve, a silicone bladder, and 3D-printed self-sealing end caps. The outer braided mesh radially shortens when pressurized, while the knit-elastic layer stretches longitudinally, producing a bending motion greater than traditional inelastic-based fabric actuators. The actuators are mounted parallel to the knee and connected to semi-rigid, anatomically conforming mounts integrated with custom neoprene pants. The reported function is generation of substantial forces sufficient to assist sit-to-stand transitions (Liu et al., 2024).

A more general actuation framework is provided by soft sleeve actuators fabricated from thermoplastic elastomers using a customized fused filament fabrication process. Four modalities are reported: linear, bending, twisting, and omnidirectional. The linear actuator delivers force output $>210\ \mathrm{N}$ with extension $>50\ \mathrm{mm}$ and contraction up to 19 mm; the bending actuator achieves angles $>140^\circ$ and force output $>38\ \mathrm{N}$; the twisting actuator reaches up to $= 97.1\%$0 clockwise and $= 97.1\%$1 counterclockwise at effective pressures $= 97.1\%$2; and the omnidirectional actuator provides bending force $= 97.1\%$3, extension $= 97.1\%$4, contraction up to 25 mm, and bending over $= 97.1\%$5. The actuators are described as self-supporting, hollow sleeve-like mechanisms that reduce the need for complex attachment hardware (Abboodi, 8 Nov 2025).

The semi-rigid approach is exemplified by Exo-Muscle, a knee assistive device that uses a semi-rigid chain mechanism to form a predefined tendon route around the knee. The chain is flexible when needed and rigidifies along a predictable path as knee flexion changes, which is reported to eliminate misalignment with the knee joint center of rotation while preserving deterministic tendon routing. The device integrates IMUs, a load cell, a Bowden cable transmission, and a TREE Lime actuator, and is reported to provide up to 38 Nm assistive torque at the knee (Zhang et al., 14 Mar 2025).

At the passive end of the spectrum, the mechanical orthosis mechanism adds a lever arm below the knee to facilitate leg extension and knee locking. Its operating principle is expressed as

$= 97.1\%$6

with the lever arm increasing torque for a given user-applied force. In the reported single-user test, extension and locking were described as “simple but effective, and less demanding,” as well as “faster and easier” (Lemire et al., 2020).

4. Learning, decoding, and closed-loop interaction

Smart leg sleeves are increasingly coupled to learning-based inference pipelines. For whole-body pose estimation from knee sleeves, the learning target is quaternion orientation of major lower-body joints. The reported loss is quaternion distance,

$= 97.1\%$7

and the evaluation metric is RMSE in angle per joint. This formulation is appropriate because the output space is rotational and supports full lower-body motion reconstruction from wearable data alone (Zhang et al., 2023).

For exoskeleton interaction, the layered smart leg sleeve uses task-specific neural decoders. Ankle joint moment estimation is based on unilateral sEMG and IMU over a 200 ms window with 10 ms stride, using a dual-branch network with a Temporal Convolutional Network for IMU and a 1D SE-ResNet for sEMG. Injury risk detection uses unilateral strain signals with a 1 s window and 50 ms stride, while metabolic trend classification uses bilateral sEMG and IMU with longer temporal windows reflecting slower metabolic dynamics. The formulation reported for joint moment estimation is

$= 97.1\%$8

This indicates a shift from sleeve-based sensing as passive measurement toward sleeve-based sensing as a real-time control substrate (Tang et al., 16 Aug 2025).

Fabric sleeves have also been used as programmable haptic interfaces for motor learning. The electroadhesive haptic sleeve uses independently controllable ventral and dorsal textile clutches with recovery time $= 97.1\%$9, holding force modulated by applied voltage according to

$= 0.96$0

In a motor learning study, the haptic group committed 23.5% fewer errors than the visual-only group during evaluation, with path-following error decreasing by 37.56% and waypoint-navigation error improving by 45.64%. The paper explicitly states that the mechanism is directly extensible to joints in the lower limb, forming a “Smart Leg Sleeve” for gait rehabilitation, fall prevention, or lower-limb teleoperation (Ramachandran et al., 2021).

The data-consistency problem links directly to learning efficacy. In the sEMG sleeve literature, the processed EMG features $= 0.96$1 are mapped by a 74-layer ResNet to predicted 6-DOF kinematics,

$= 0.96$2

Using consistent recordings accumulated over time, the system demonstrated simultaneous and proportional control of six degrees of freedom even 263 days after initial algorithm training. A plausible implication is that lower-limb smart sleeves may benefit similarly from long-horizon, session-consistent datasets (George et al., 2020).

5. Applications across rehabilitation, mobility assistance, and human-exoskeleton interaction

The most immediate application domain is rehabilitation and mobility support. The hyper-bending knee-extension exosuit is explicitly designed for people with impaired mobility during demanding movements such as sit-to-stand transitions, while the soft sleeve actuator work is framed around wearable mobility assistive devices for aging populations and neurological and musculoskeletal disorders (Liu et al., 2024, Abboodi, 8 Nov 2025).

Motion assessment and remote monitoring form a second major application cluster. The Intelligent Knee Sleeves are presented for healthcare, sports, home fitness, elderly care, and physical rehabilitation, including monitoring of physiotherapy for knee injuries, osteoarthritis, and post-surgical rehabilitation. Their wearable sensing avoids line-of-sight requirements and camera susceptibility to occlusion, lighting, reflective surfaces, and ambient visual noise, enabling use in cluttered environments and outdoor settings (Zhang et al., 2023).

Human-exoskeleton interaction is a third application axis. The layered smart leg sleeve is designed to support three core objectives: controlling personalized assistance, optimizing user effort, and safeguarding against injury risks. Its integration with a custom soft ankle exoskeleton, combined with low total sensor-and-electronics weight of $= 0.96$3, places the sleeve within the emerging category of physiologically informed exoskeleton interfaces rather than stand-alone wearables (Tang et al., 16 Aug 2025).

Finally, sleeve-form systems can support motor training rather than direct assistance. The electroadhesive haptic sleeve demonstrates that physically constraining erroneous movement can improve retention and transfer better than visual feedback alone in a drone-control task. Since the paper explicitly identifies extension to lower-limb joints, this provides a concrete template for smart leg sleeves that teach rather than merely measure or actuate (Ramachandran et al., 2021).

6. Design trade-offs, misconceptions, and research directions

A recurrent design tension concerns rigidity versus compliance. Rigid exoskeletons provide deterministic torque but can create misalignment because the human knee does not behave as a fixed-axis hinge, whereas soft exosuits adapt better to anatomy but may suffer from unpredictable tendon routing and less deterministic load compensation. Semi-rigid mechanisms such as Exo-Muscle explicitly position themselves between these two extremes, attempting to retain adaptability while enforcing a well-defined tendon path (Zhang et al., 14 Mar 2025).

A second misconception is that sleeve-form devices are limited to low-authority sensing. The current literature includes sleeves that estimate full lower-body motion, infer ankle joint moment and metabolic trends, physically constrain movement with electroadhesive clutches, and generate forces and torques sufficient for knee assistance. The available evidence therefore supports a broader interpretation of the smart leg sleeve as a platform for perception, control, and force delivery rather than as a garment with embedded sensors alone (Zhang et al., 2023, Tang et al., 16 Aug 2025, Ramachandran et al., 2021, Liu et al., 2024).

Future directions are increasingly shaped by adjacent work in compliant leg robotics. Research on continuously compliant robotic legs argues for lower mass, simplified actuation design, and direct force control through deformable structures, and explicitly notes relevance for future exosuits and wearable designs (Bendfeld et al., 2024). Work on the DecARt Leg similarly suggests that reduced distal inertia, decoupled actuation, and actuator placement above the knee may inspire future wearable robots or assistive prosthetics such as smart leg sleeves (Davydenko et al., 13 Nov 2025). This suggests that future smart leg sleeves may not converge on a single textile-only paradigm; instead, they may combine smart textiles, distributed compliance, semi-rigid force routing, and proximal actuation in anatomically integrated lower-limb systems.