A laser-engraved wearable gait recognition sensor system for exoskeleton robots

Exoskeleton robots have seen wide application in freight transport, disaster relief, and rehabilitation due to their ability to enhance or extend human capabilities. For lower-limb assisted exoskeletons, motion cannot be preprogrammed solely from environmental context; instead, it must follow the wearer’s intention and assistance requirements, where gait information is crucial for motion planning, assistance strategy, and evaluating assistance effects. Accurate real-time gait recognition is thus a key task for human–robot cooperation in lower-limb exoskeletons. Wearable sensors capturing plantar pressure, velocity, and acceleration are suitable for gait perception, but systems must be portable, compliant, and comfortable. While flexible electronics have enabled lightweight, comfortable sensors, batch-to-batch variability and fabrication limitations hinder real-life deployment. Customized laser direct writing offers efficient, precise, and environmentally friendly fabrication of sensitive materials and micro/nanostructures for flexible sensors. Laser-induced graphene (LIG), produced by one-step laser ablation of polyimide (PI), serves as a versatile sensing material and electrode, and can be integrated with soft polymers such as PDMS to form flexible composites. In this work, the authors design a wearable LIG-based insole sensor system for exoskeleton robots to achieve real-time gait recognition, combining laser-fabricated LIG/PDMS pressure sensor units and interdigital electrodes with supporting hardware and machine learning algorithms. The system aims to provide high-accuracy gait information for exoskeleton control, with potential applications in gait analysis and rehabilitation medicine.

The paper situates its contribution within several areas: (1) Wearable gait sensing for exoskeletons, where capturing plantar pressure and related foot dynamics informs motion intention and control strategies. (2) Advances in flexible electronics enabling lightweight, comfortable sensors (pressure, temperature, humidity, strain, biosensors) for human activity tracking, though often limited by batch-to-batch variability from fabrication constraints. (3) Customized laser direct writing, an efficient and precise green manufacturing method to generate functional micro/nanostructures and sensitive materials directly, which has yielded various flexible sensors. (4) Laser-induced graphene (LIG) produced via one-step laser ablation of polyimide films, used both as sensing elements and interconnection electrodes, and transferable onto soft matrices (e.g., PDMS, ecoflex, hydrogels) to create flexible composites. These developments motivate a laser-fabricated LIG-based plantar pressure system to address variability and integration challenges in practical gait recognition for exoskeleton control.

Sensor design and fabrication: The pressure sensor comprises three layers: (i) a PI film with patterned LIG forming interdigital electrodes, (ii) a laser-textured LIG/PDMS sensing layer, and (iii) a PET encapsulation layer. Fabrication steps include: (1) forming the LIG-sensitive layer and interdigital electrodes via laser ablation on PI films; (2) transferring the LIG-sensitive layer onto soft PDMS; (3) laser texturing the LIG/PDMS to create hierarchical microstructures; and (4) aligning and assembling the LIG/PDMS sensing unit with the interdigital LIG electrodes and PET encapsulation. Sensing mechanism: Pressure-induced changes in the effective contact area between the laser-textured LIG/PDMS surface and the interdigital LIG electrodes modulate the electric current. Sensitivity is defined as S = (ΔI/I0)/P, where I0 is the baseline current without load, P is applied pressure, and ΔI is the change in current under pressure. Materials and characterization: LIG morphology was examined via SEM; Raman spectra showed characteristic D (~1350 cm−1), G (~1580 cm−1), and 2D (~2700 cm−1) peaks. Laser texturing produced hierarchical microstructures with ~40 µm depth, enhancing contact junction formation under load. Performance optimization: The effects of laser texturing and laser fluence on sensitivity and baseline resistance were studied. For fluence < 8 J/cm², resistance remained unchanged, indicating retained LIG interconnectivity; higher fluence increased resistance substantially. Devices fabricated at 6.2 J/cm² exhibited the highest sensitivity, balancing structure depth and effective contact area. Cycling and dynamic tests included repeated loading at 50–300 kPa and frequencies of 0.05–0.2 Hz to assess stability. Insole system integration: Multiple LIG pressure sensor units (seven total) were integrated into a flexible, foldable intelligent insole (255 mm × 145 mm × 0.15 mm) positioned at plantar stress points. Three key plantar sampling regions (CH1, CH2, CH3) were defined based on plantar pressure distribution. A 9-channel terminal connects the insole to a hardware module via ribbon cables. Hardware and signal processing: The electronics include an input signal amplifier (ISA), analog-to-digital converter (ADC), and microcontrol unit (MCU). Data communication among modules uses a controller area network (CAN). The insole is embedded in the exoskeleton shoe and connected mechanically to the shank rod. Gait recognition and control: The system targets real-time identification of gait phases—initial contact (IC), loading response (LR), mid-stance (MS), terminal stance (TS), and swing (SW). The data pipeline involves normalization, scaling, model training, and testing to classify gait phases from multi-channel plantar pressure signals. Recognized gait phases inform the exoskeleton’s human–robot interaction control and trajectory planning.

- The LIG-based intelligent insole with multiple pressure sensor units accurately recognizes gait phases for exoskeleton control, achieving 99.85% gait recognition accuracy in experiments. - Laser texturing of the LIG/PDMS sensing surface significantly increases sensitivity compared to untextured LIG/PDMS by increasing contact junctions with interdigital electrodes. - Optimal laser fluence for fabricating the sensing microstructures was 6.2 J/cm², yielding the highest sensitivity; for fluence < 8 J/cm², baseline resistance remained unchanged, whereas higher fluence increased resistance up to approximately 546.4%. - Hierarchical microstructures with ~40 µm depth were achieved on LIG/PDMS, contributing to enhanced pressure responsiveness. - The sensors demonstrated robust cycling stability with negligible degradation over ~5000 loading cycles under 300 kPa; stable performance was also observed at 0.05, 0.1, and 0.2 Hz and across pressures of 50, 100, and 200 kPa. - The insole is thin and flexible (255 mm × 145 mm × 0.15 mm) and integrates seven sensor units with three principal sampling regions (CH1–CH3) and a 9-channel interface for system integration. - The system was validated on an exoskeleton robot, confirming effective real-time plantar pressure acquisition and gait phase recognition for human–robot interaction.

The study addresses the need for accurate, real-time gait perception in lower-limb exoskeletons by integrating a flexible LIG-based plantar pressure array with dedicated electronics and a machine learning classifier for gait phase recognition. The laser-direct-write approach enables consistent fabrication of sensitive LIG structures directly on PI, with subsequent transfer and laser texturing on PDMS to enhance sensitivity through increased contact junction dynamics. The resulting insole maintains comfort and compliance for wearability while providing high-fidelity pressure signals. Performance optimization via laser fluence selection (notably at 6.2 J/cm²) and texturing yields strong sensitivity without compromising interconnectivity, and long-term cycling demonstrates durability suitable for practical use. The high recognition accuracy (99.85%) and successful integration within an exoskeleton control loop indicate that the system can reliably inform trajectory planning and assistance strategies across the gait cycle phases (IC, LR, MS, TS, SW). These results underscore the relevance of LIG-based flexible sensors and tailored microstructures for robust human–robot interaction in assistive robotics and rehabilitation contexts.

The work presents a wearable LIG-based plantar pressure sensor insole with integrated hardware and a machine learning algorithm for real-time gait recognition in exoskeleton robots. Using laser direct writing and laser texturing, the authors fabricated sensitive LIG/PDMS microstructured sensors coupled with interdigital electrodes on PI and PET encapsulation. The system achieves high gait recognition accuracy (99.85%), exhibits strong durability over 5000 cycles, and integrates seamlessly with an exoskeleton for control applications. The intelligent insole demonstrates strong potential for gait analysis and recognition in rehabilitation medicine and for enhancing human–robot interaction in assistive exoskeletons.