ETD

• assistance training
• heart rate
• hip exoskeleton
• lower limb
• optimization
• resistance training
• torque estimation

Cardiovascular and cardiorespiratory system disorders are known to be the leading cause of disease and mortality worldwide, and are often associated with aspects such as aging and lifestyle habits. Aging is accompanied by a progressive decline in both musculoskeletal and cardiovascular function, which significantly contributes to functional impairment in older adults. Physical activity plays a crucial role in reducing the risk of having these disordes, however, sufficient engagement remains a global challenge. A recent research area has investigated the use of wearable robotic devices for resistance training purposes. To achieve adaptability, physiological signals have been included in these systems to enable a more personalized assistance, potentially enhancing both performance and the overall exercise experience. This work presents the design, implementation, and validation of an open-loop cardiovascular regulation strategy applied to a bilateral hip exoskeleton (APO-R). The objective was to modulate the user’s heart rate response during treadmill walking by adjusting the level and mode of exoskeleton actuation, as well as to evaluate the corresponding physical effort under different walking speeds.

To develop and validate the proposed system, two main experimental stages were conducted. In the first stage, four healthy subjects participated in a four-day protocol designed for familiarization with the hip exoskeleton and evaluation under five actuation conditions: high and low assistance modes (H-AM, L-AM), high and low resistance modes (H-RM, L-RM), and transparent mode (TM). These were tested at three walking speeds (slow, normal, and fast). From these trials, the Physiological Cost Index (PCI) and Metabolic Cost of Transport (MCoT) were computed and analyzed to quantify the physical effort associated with each condition.

In the second stage, an energy-based optimization algorithm was implemented using Sequential Quadratic Programming (SQP) to determine the torque parameters required to achieve predefined energy levels at each step. Four healthy subjects participated in a single-day session, testing four conditions (NoExo, TM, H-AM, and H-RM) across the same three walking speeds. System performance was assessed based on the accuracy of the achieved energy values and the influence on the PCI under each actuation mode.

From the first stage, results showed that physiological measures (PCI and MCoT), exhibited similar behavioral trends, suggesting that both can serve as representative indicators of physical effort under exoskeleton actuation. From the second stage, the results show that the SQP algorithm successfully estimated the parameters required to reach target energy levels, achieving accuracies above 86% in assistance and 90% in resistance conditions.