TL;DR
This paper demonstrates that a hip exoskeleton, originally designed for walking, can significantly reduce the metabolic cost of cycling by up to 31.4%. By optimizing the timing of the exoskeleton's assistance using a human-in-the-loop approach, researchers achieved substantial energy savings, highlighting the potential for exoskeletons to enhance cycling performance and accessibility.
What Problem Does This Paper Address?
As individuals age or experience certain health conditions, their physical capabilities, including endurance and strength, can decline. This decline can impact their ability to participate in activities like cycling, which is a popular form of exercise, rehabilitation, and transportation. While electric bikes offer assistance, autonomous exoskeletons for lower limbs have shown promise in reducing the effort required for walking. However, the application of such exoskeletons to cycling, specifically to reduce the metabolic cost of pedaling, has not been extensively explored. This paper aims to bridge this gap by investigating the feasibility and effectiveness of using a hip exoskeleton, designed for walking, to assist cyclists and reduce their energy expenditure.
Methods
The researchers utilized the GuroX hip exoskeleton, which was originally developed for walking assistance. For this study, the exoskeleton was adapted to provide assistance during cycling on a cycle ergometer. Two primary methods were employed to determine the optimal assistance strategy:
- Sweep Protocol: This method involved systematically varying the timing of the exoskeleton's peak extension and flexion torques across a range of pedal stroke percentages. This allowed for a broad exploration of different assistance timings.
- Human-in-the-Loop (HITL) Optimization: This technique used the participants' metabolic cost as feedback to iteratively refine the exoskeleton's assistance timing. A Bayesian optimization-based method was employed to efficiently search for the optimal parameters that minimized metabolic cost.
Metabolic cost was measured using indirect calorimetry, and crank kinematics were captured using motion capture. The exoskeleton's torque and power were calculated based on motor current and gear ratios.
Datasets
Not discussed in the paper. The study involved data collection from three human participants during controlled cycling experiments.
Key Insights and Ideas
- Repurposing Walking Exoskeletons for Cycling: The study successfully demonstrated that a hip exoskeleton designed for walking can be effectively adapted to provide meaningful assistance during cycling, reducing metabolic cost. This suggests a broader applicability of lower-limb exoskeletons beyond their initial intended use.
- Timing is Crucial for Assistance: The research strongly indicates that the timing of the exoskeleton's assistance is a critical factor in maximizing its effectiveness. Simply applying torque isn't enough; it needs to be synchronized with the natural rhythm of the pedal stroke.
- HITL Optimization for Personalization: The human-in-the-loop optimization approach proved valuable in identifying individualized assistance strategies that further reduced metabolic cost, underscoring the need for personalized solutions in exoskeleton control.
- Beyond Motor Power: While increased motor power from the exoskeleton is a primary driver of reduced effort, participant feedback highlighted the importance of user comfort and the synchronization between the user's movements and the exoskeleton's actions. This suggests that a holistic approach to exoskeleton design and control is necessary.
Key Results
- Significant Metabolic Cost Reduction: The GuroX hip exoskeleton, with a peak assistance torque of approximately 10.3 Nm, substantially reduced the net metabolic cost of cycling. Reductions of 31.4 ± 8.1% were observed compared to the transparent (no assistance) condition and 26.4 ± 14.1% compared to cycling without the exoskeleton.
- Optimal Assistance Timing: The optimal timing for peak extension torque was found to be around 15.3% of the pedal stroke, and for peak flexion torque, it was around 65.3% (a 50% offset from extension).
- Motor Power Contribution: The average motor power provided by the exoskeleton was a significant contributor to the reduced cycling effort. The "Optimal" condition, identified through HITL, provided slightly higher average motor power (13.9 W) compared to the minimum metabolic cost condition from the sweep session.
- Participant Variability: While the group achieved significant reductions, individual results showed some variation, reinforcing the idea that personalized assistance strategies are beneficial.
Limitations and Future Work
- Small Sample Size: The study was conducted with only three unimpaired participants. This limits the generalizability of the findings, and a larger and more diverse participant pool is needed for validation.
- Lab Environment: The experiments were performed in a controlled laboratory setting using a cycle ergometer. Future research should validate these findings in real-world cycling conditions.
- Comfort and Synchronization: While motor power was identified as a key factor, participant feedback pointed to the importance of user comfort and the synchronization between the user and the exoskeleton. Future work should focus on enhancing the comfort of human-machine interfaces to allow for higher torques and better integration.
- Parameter Exploration: The authors suggest exploring a wider range of parameters for HITL optimization to further enhance individualized assistance strategies.
- Muscle Activity Analysis: Investigating changes in muscle activity could provide deeper insights into the specific biomechanical contributions to the decreased effort during assisted cycling.
Summary updated