A series-elastic actuator (SEA) puts a spring in line with the ankle joint. As the body's weight advances over a planted foot (controlled dorsiflexion, roughly 15–45% of the gait cycle), the ankle rotates the spring and it stores elastic energy instead of the calf muscles absorbing it eccentrically:
Spring torque: τ_s = k·Δθ
Stored energy: E = ½·k·Δθ² (Δθ = dorsiflexion angle, radians)
Near push-off (45–62%) the ankle rapidly plantarflexes and the same spring unloads, returning that energy as assistive torque — literally the mechanism the article describes as a "passive" exoskeleton, distinct from an actively powered one. A hybrid unit adds a motor that contributes a further fraction of the required torque on top of the spring:
τ_required(φ) — biomechanical demand at gait phase φ
τ_assist(φ) = τ_spring_release(φ) + gain·τ_required(φ)
τ_biological = max(0, τ_required(φ) − τ_assist(φ))
W_bio = Σ τ_biological·|Δθ| over one full stride
The "Patient work / stride" readout integrates τ_biological over one full gait cycle — the model's proxy for the calf muscles' own metabolic effort. Raising k stores/returns more energy during the amber "loading" arc and cyan "releasing" arc on the phase ring; raising the motor gain shaves torque off the whole cycle, not just the elastic window. Switching to "Rigid" sets both to zero — a stiff, unassisted boot — so the reduction percentage always compares against that unassisted baseline.
This is a simplified educational model (idealised joint trajectory, no muscle-tendon dynamics) built to make the store-and-release principle of passive/hybrid exoskeletons visible and adjustable, not a clinical prediction tool.