Researchers at Hainan University have developed a novel approach to modeling gait synergy in lower limb assistive devices, potentially revolutionizing the control of powered prostheses and exoskeletons. The study, published in Cyborg and Bionic Systems, introduces a two-stage strategy called FS-Seq2Seq that outperforms existing methods in generating user-adaptive and synergic trajectories.
Understanding Gait Synergy and Its Importance
Gait synergy refers to the coordinated movement patterns between different parts of the body during walking. For individuals using lower limb assistive devices, such as prosthetic legs or exoskeletons, replicating this natural synergy is crucial for comfortable and efficient movement.
Professor Fengyan Liang, lead author of the study, explains: “Control of active assistive devices is a critical and challenging issue when designing and generating user-, temporal-, and phase-adaptive and synergistic reference trajectories for various patients.”
The concept of gait synergy allows assistive devices to predict appropriate movements for affected or missing limbs based on the motion of healthy body parts. This approach aims to create a more natural and intuitive human-machine interface.
FS-Seq2Seq: A Two-Stage Approach to Better Movement
The research team’s new method, FS-Seq2Seq, combines two key elements:
- Feature Selection (FS): This process identifies the most relevant input data, reducing noise and improving model efficiency.
- Sequence-to-Sequence (Seq2Seq) modeling: An advanced neural network architecture that excels at translating one sequence of data into another.
The study compared Seq2Seq with other popular neural network models, including Long Short-Term Memory (LSTM), Recurrent Neural Networks (RNN), and Gated Recurrent Units (GRU). Seq2Seq consistently outperformed these alternatives in both interlimb (between different limbs) and intralimb (within the same limb) synergy modeling.
Key findings include:
- Seq2Seq achieved a mean absolute error of 0.404° for interlimb synergy and 0.596° for intralimb synergy, surpassing other models.
- Feature Selection significantly improved Seq2Seq’s performance (p < 0.05).
- FS-Seq2Seq outperformed methods used in existing studies.
Why It Matters: Improving Quality of Life for Device Users
This research has significant implications for individuals who rely on lower limb assistive devices:
- More Natural Movement: By better replicating natural gait synergy, devices can provide more intuitive and comfortable assistance.
- Improved Adaptability: The FS-Seq2Seq approach allows for user-specific customization, potentially accommodating a wider range of mobility needs.
- Enhanced Human-Machine Interaction: More accurate prediction of movement intentions could lead to smoother and more responsive device control.
Professor Liang emphasizes the broader impact: “This study emphasizes the promise of synergy-based trajectory prediction for assistive devices and provides insights into achieving optimal modeling with optimal feature combinations, resulting in synergic and user-adaptive trajectories that improve human-machine interactions.”
The research also highlights the importance of systematic feature selection in future studies, potentially leading to further improvements in assistive device control.
While the results are promising, questions remain about real-world implementation:
- How will the FS-Seq2Seq method perform across diverse patient populations?
- What computational resources are required to run this model in real-time on assistive devices?
- How might this approach be integrated with existing assistive technologies?
As research in this field continues, the FS-Seq2Seq method offers a compelling new direction for improving the lives of those who rely on lower limb assistive devices. Future studies will likely focus on refining these techniques and addressing the challenges of practical implementation across various device types and user needs.
