Wearable technologies for assisted mobility in the real world

Shuo Gao, Jianan Chen, Yunjia Xia, Xuemeng Li, Weihao Ma, Huixin Yang, Jinchen Li, Xinkai Zhou, Tianyu Jia, Yuchen Xu, Julie Uchitel, Dean Ta, Peng Qi, Junbo Ge, Yi Guo, Yajie Qin, Inseung Kang, Wenyao Xu, He Li, Jinke Chang, Siming Zuo, Shiwei Wang, Shan Luo, Letizia Gionfrida, Chen Hu, Shuqin Dong, Yong‐Xin Guo, Yixuan Yuan, Haixia Zhang, Haotian Chen, Yu Pan, Chenyun Dai, Qinyuan Ren, Rui Loureiro, Tom Carlson, Wei Chen, Yuan‐Ting Zhang, P. A. Kyriacou, Hadi Heidari, Kianoush Nazarpour, Themis Prodromakis, Alexander J. Casson, Tamar R. Makin, Gert Cauwenberghs, Dario Farina, Hubin ZhaoNature Communications 2025 · DOI

TL;DR

This perspective paper provides a comprehensive overview of wearable technologies for assisted mobility, highlighting the integration of neural sensing, electronics, and co-design to create personalized and responsive solutions. It emphasizes the critical role of multimodal fusion and human-in-the-loop strategies in developing intelligent systems that enhance autonomy and inclusion for individuals with mobility impairments.

What Problem Does This Paper Address?

Approximately 1.3 billion people worldwide live with significant disabilities, and around 58 million experience mobility impairments. While traditional assistive devices have improved mobility, they often lack the flexibility and adaptability needed for dynamic, real-world environments. This paper addresses the challenge of creating assistive mobility solutions that are not only functional but also personalized, responsive, and seamlessly integrated into users' lives, thereby restoring function and enhancing independence.

Methods

The paper focuses on wearable sensing technologies and their multimodal fusion for assisted mobility. Key sensing modalities discussed include:

  • Central Interfacing:
    • Electroencephalography (EEG): Captures brain activity for intent recognition, particularly through motor imagery (MI) and steady-state visual evoked potentials (SSVEP).
  • Functional Near-Infrared Spectroscopy (fNIRS): Monitors cortical activity by detecting changes in oxygenated and deoxygenated hemoglobin, offering better spatial resolution than EEG.
  • Peripheral Interfacing:
    • Electromyography (EMG): Detects electrical signals from muscles to infer movement intent.
  • Electrooculography (EOG) & Eye-tracking: Monitors eye movements for directional control.
  • Inertial Measurement Units (IMUs): Track body motion, acceleration, and orientation for gait analysis and balance control.
  • Multimodal Fusion: The paper emphasizes combining these diverse sensing modalities (e.g., EEG-EMG, EEG-fNIRS) to overcome individual limitations and improve accuracy, responsiveness, and robustness. Different fusion strategies (data-level, feature-level, decision-level) are discussed.
  • Environmental Perception: Technologies like computer vision (CV) and depth cameras are highlighted for enabling obstacle detection and avoidance, crucial for safe navigation.
  • Human-in-the-Loop & Co-design: The importance of user involvement in the design process and adaptive control systems that integrate user feedback is stressed.

Datasets

Not discussed in the paper.

Key Insights and Ideas

  • The Shift to Personalized, Adaptive Systems: The core idea is moving beyond one-size-fits-all assistive devices to highly personalized systems that adapt in real-time to the user's physiological and environmental context.
  • Multimodal Fusion is Key: No single sensor modality is perfect. Combining signals from the brain, muscles, and body motion offers a more comprehensive and reliable understanding of user intent and state.
  • Environmental Awareness is Crucial: For true mobility assistance, devices need to "see" and understand the environment to navigate safely and effectively.
  • Human-Centric Design is Non-Negotiable: For these technologies to be adopted and effective, they must be developed with direct user input, ensuring usability, comfort, and alignment with real-world needs.
  • The "Uncanny Valley" of Assistive Tech: Just as in AI and robotics, assistive devices that approximate but don't perfectly replicate natural human function can lead to user discomfort and reduced adoption.

Key Results

The paper is a perspective piece, so it doesn't present new experimental results. Instead, it synthesizes existing research to highlight:

  • Advancements in Sensing: Significant progress has been made in miniaturizing and improving the accuracy and reliability of wearable sensors like EEG, fNIRS, EMG, and IMUs.
  • Successful Integration: Numerous studies demonstrate the successful integration of these sensors into various assistive devices, including exoskeletons, prosthetics, smart wheelchairs, and non-visual navigation aids.
  • Improved Control and Responsiveness: Multimodal fusion and adaptive control strategies have shown promise in enhancing the intuitiveness and responsiveness of these devices.
  • Growing Potential for Independence: The collective advancements point towards a future where individuals with mobility impairments can achieve greater autonomy and participate more fully in society.

Limitations and Future Work

The authors acknowledge several limitations and suggest future directions:

  • Technical Limitations:
    • Sensor Noise and Artifacts: Wearable sensors are susceptible to noise and motion artifacts, impacting signal reliability.
  • Limited Battery Life: Continuous operation of complex wearable systems requires significant power, posing a challenge for long-term use.
  • Spatial and Temporal Resolution Trade-offs: Different sensing modalities have inherent limitations in spatial or temporal resolution.
  • Signal Processing Delays: Delays in sensing, processing, and actuation can disrupt seamless integration and user experience.
  • User-Centric Challenges:
    • User Customization and Training: Many systems require extensive user-specific calibration and training, hindering widespread adoption.
  • Cognitive Load: Integrating complex wearable interfaces can impose a significant cognitive burden on users, especially during multitasking.
  • Cost and Accessibility: High costs and limited availability of advanced assistive technologies remain barriers for many.
  • Privacy Concerns: The collection of sensitive physiological and behavioral data raises privacy concerns.
  • Future Work:
    • Developing Robust Fusion Algorithms: More sophisticated and generalized fusion techniques are needed.
  • Enhancing Environmental Perception: Advancements in AI and computer vision are crucial for better environmental understanding.
  • Improving Human-Machine Coordination: Deeper understanding of sensorimotor-cognitive integration is required.
  • Soft Robotics and Biomaterials: Development of more comfortable and adaptable soft exoskeletons and prosthetics.
  • Standardized Evaluation Frameworks: A need for unified methods to assess the effectiveness, usability, and long-term impact of assistive technologies.
  • Co-creation and Participatory Design: Continued emphasis on involving users throughout the development lifecycle.
  • Addressing Affordability and Inclusivity: Making these technologies more accessible and affordable for diverse populations.

Summary updated

Abstract

Mobility impairments from aging, injury, or medical conditions limit independence and social participation. Conventional assistive devices lack adaptability in complex environments. Recent wearable technologies integrating neural sensing, electronics, and co-design offer personalized, responsive mobility support. This perspective focuses on advances in wearable sensing and multimodal fusion for intent recognition, environmental interaction, and adaptive control in exoskeletons, prosthetics, smart wheelchairs, and navigation systems. Emphasizing human-in-the-loop and cognitive-sensorimotor integration, it outlines emerging trends and challenges, promoting intelligent, user-centered solutions to restore function and enhance autonomy, accessibility, and inclusion for individuals with mobility impairments.
BibTeX
@inproceedings{gao2025wearable,
  author = {Shuo Gao and Jianan Chen and Yunjia Xia and Xuemeng Li and Weihao Ma and Huixin Yang and Jinchen Li and Xinkai Zhou and Tianyu Jia and Yuchen Xu and Julie Uchitel and Dean Ta and Peng Qi and Junbo Ge and Yi Guo and Yajie Qin and Inseung Kang and Wenyao Xu and He Li and Jinke Chang and Siming Zuo and Shiwei Wang and Shan Luo and Letizia Gionfrida and Chen Hu and Shuqin Dong and Yong‐Xin Guo and Yixuan Yuan and Haixia Zhang and Haotian Chen and Yu Pan and Chenyun Dai and Qinyuan Ren and Rui Loureiro and Tom Carlson and Wei Chen and Yuan‐Ting Zhang and P. A. Kyriacou and Hadi Heidari and Kianoush Nazarpour and Themis Prodromakis and Alexander J. Casson and Tamar R. Makin and Gert Cauwenberghs and Dario Farina and Hubin Zhao},
  title = {{Wearable technologies for assisted mobility in the real world}},
  booktitle = {Nature Communications},
  year = {2025},
  doi = {10.1038/s41467-025-67126-4},
}
Figure