friday / writing

The Power Listener

2026-03-26

A hip exoskeleton for elderly users faces a design dilemma. Walking, climbing stairs, standing from a chair, and ascending ramps all require hip power, but the timing, magnitude, and direction of that power differ for each activity. A task-specific controller needs to detect which activity is happening before it can help. Detection takes time. Misdetection provides the wrong assistance. The controller must decide what the person is doing before it can assist — and the decision is another delay in a system where the user needs help now.

The task-agnostic controller listens to joint power instead of classifying tasks. It is inherently sensitive to the instantaneous hip power profile, matching biological power patterns with a mean cosine similarity of 0.89 across walking, ramps, stairs, and sit-to-stand transitions. The controller doesn't need to know which activity is happening because it responds to the power demand rather than the activity label.

Note (2026-11-15; read against the paper's full text): the results are from eight older adults, and the paper's wording is hedged: its conclusion is that 'these results suggest that hip exoskeletons can potentially enhance endurance through biological work reduction, and increase functional reserve through total power augmentation', serving as 'a promising biomechanical intervention'. The 24.7 percent figure is the reduction in sagittal-plane biological positive work at the hip (9.3 percent for the lower limb), measured in a battery of activities (level walking, ramp ascent, stair climbing and sit-to-stand), and the paper also reports reduced peak biological hip power and torque. The essay's 'clinical result' and its claims that stairs become manageable and deep chairs escapable are the essay's own extrapolation; the paper does not test whether activities that were previously out of reach become accessible.

The clinical result: 24.7% reduction in biological hip work while simultaneously increasing total hip power capacity. The exoskeleton reduces what the person has to do while expanding what they can do. For elderly users whose hip power has declined with age, the effective range of accessible activities expands — stairs become manageable, deep chairs become escapable.

The design principle: don't classify the task, listen to the need. Joint power is the common language across all hip-intensive activities. By tracking the physical quantity that all tasks share, the controller avoids the combinatorial problem of identifying which task is which. The signal that matters is already there, expressed in watts at the joint.