A More Efficient Wearable Hip Exoskeleton: Moving Beyond Rigid Gait Algorithms

Engineers have successfully programmed a wearable hip exoskeleton that adapts to human movement without needing to predict exactly where a user is in their walking cycle. This was historically difficult because human walking is highly variable; older algorithms failed the moment a person turned a corner or changed their stride.

The Wearable Hip Exoskeleton Problem

Conventional assistive devices rely on explicit gait phase estimation. They use pre-defined torque profiles synced to a rigid step cycle. If you walk in a straight line at a constant speed, the motors assist you efficiently. However, if you turn or stumble, the rigid profile fights your natural movement. The previous alternative, Delayed Output Feedback Control (DOFC), stopped predicting steps and instead used delayed joint angles to apply force. DOFC prevented the exoskeleton from fighting the user, but it could only produce very basic force patterns, limiting its overall efficiency.

Adding Harmonics to the Algorithm

To fix this, researchers introduced Modified Delayed Output Feedback Control (MDOFC). They added harmonic components to the algorithm, allowing the device to generate highly adaptable torque profiles without reverting to rigid step predictions. The team optimised these parameters using a neural network and forward dynamic gait simulations. The computational simulations yielded promising results, showing that MDOFC successfully maintained dynamic stability up to 45 watts. Moving from simulation to human trials, the researchers measured two specific physical outcomes:

• Comparable net metabolic cost reduction in users.
• An 8.8 per cent drop in mechanical power consumption compared to the older DOFC method.

The human experiments confirmed that MDOFC provides the same metabolic relief to the wearer as previous models while demanding significantly less mechanical effort from the machine.

What the Data Leaves Unanswered

This reduction in mechanical power is highly practical, as it suggests future devices could operate using smaller batteries and lighter motors. Yet, the study leaves several operational challenges unresolved. The current data relies on simulations and early-stage human testing limited to controlled laboratory conditions. It does not solve the broader issue of complex terrain variability. We do not yet know how the MDOFC algorithm handles stairs, steep inclines, or sudden slips on uneven pavements. While the mathematical framework is sound, proving its reliability across daily, unpredictable human environments remains the next necessary hurdle.