Why this problem matters

Early rehabilitation can require repetitive joint mobilization before a patient is ready for standing or walking. Conventional electric actuation can provide accurate motion, but may add mass, rigid transmission, and mechanical complexity close to the body.

SMA wires offer a different design space: high force-to-weight ratio, silent operation, and muscle-like contraction. Their use is challenging because heating and cooling introduce hysteresis, changing dynamics, and the possibility of thermal accumulation. The project therefore treats thermal recovery and actuator coordination as central control requirements.

System architecture

Research workflow

The validation strategy progresses from component behavior to system endurance:

1. Establish mechanical pretension and usable angular response.
2. Characterize relaxed and activated actuator behavior.
3. Evaluate cyclic consistency under repetitive reference tracking.
4. Test long-duration operation and analyze performance batch by batch.
5. Separate tracking evidence from safety claims and clinical interpretation.

Current evidence

The latest experimental suite evaluates global tracking error, signed bias, maximum error, state occupancy, switching behavior, and long-duration stability. In the updated master’s-thesis analysis of the 40-minute endurance experiment, the system maintained a global RMSE of approximately 0.220° over the final active analysis window.

These results support the feasibility of the control architecture on the current laboratory prototype. They do not establish clinical effectiveness or safety for patient use.

The earlier Bilbao publication used the analysis window and processing available when the conference paper was prepared. Its reported values therefore remain distinct from these updated thesis metrics and should not be treated as interchangeable.

Research context

The work is developed at Universidad Carlos III de Madrid within the RoboticsLab research environment. It is part of the context of Robótica blanda para la rehabilitación del tobillo (SRAR), R&D grant PID2023-149141OB-I00, funded by MCIN/AEI/10.13039/501100011033 and FEDER/EU.

The project page intentionally documents my research contribution and prototype development rather than serving as an official SRAR project page.

Related outputs

Next steps

• Extend sensing while preserving a clear safety hierarchy.
• Characterize thermal behavior more directly.
• Evaluate broader motion profiles and participant-specific fitting.
• Progress from engineering validation toward appropriately approved human studies.

Publication record

Authors: Sergio Jácobo-Zavaleta, Dorin Copaci, Sofía Die-Pancorbo, Dolores Blanco, and Luis Moreno
Proceedings: Simposios del Comité Español de Automática (CEA), 2(2)
Published: May 2026
DOI: 10.64117/simposioscea.v2i2.208

Overview

The paper presents an indirect thermal-aware supervision strategy for a pair of antagonistic Shape Memory Alloy actuators integrated into a modular ankle rehabilitation platform. Because SMA contraction depends on heating and recovery, accurate angular tracking alone is not sufficient: the controller must also coordinate which actuator can be energized, which actuator should recover, and when the system should remain idle.

The proposed architecture combines a bilinear PID controller with a state manager that coordinates extension, flexion, and idle operation. This supervisory layer provides a practical way to incorporate thermal recovery into control decisions without depending on direct temperature feedback in the reported setup.

Contribution

• Agonist-antagonist SMA actuation applied to ankle mobilization.
• Indirect thermal-aware supervision through state-dependent activation and recovery.
• Experimental tracking and long-duration validation on the physical prototype.
• A control architecture designed around the limitations of the actuator rather than treating them as external disturbances.

Experimental perspective

The conference paper reports a global RMSE of 0.218° and MAE of 0.163°, calculated using the analysis window and processing available when the manuscript was prepared.

The work establishes a foundation for the larger validation framework now used in the master’s thesis. Subsequent analysis separates global tracking metrics, cyclic consistency, long-duration behavior, batch-wise trends, and controller-state statistics more explicitly. In the updated thesis analysis, the corresponding endurance dataset yields 0.220° RMSE and 0.167° MAE under its revised active-window definition.

These values come from different analysis stages. The thesis metrics update the research record but do not replace the results published in the Bilbao paper.

The reported evidence concerns the engineering prototype. It should not be interpreted as clinical validation.

Research context

This research forms part of Robótica blanda para la rehabilitación del tobillo (SRAR), R&D grant PID2023-149141OB-I00, funded by MCIN/AEI/10.13039/501100011033 and FEDER/EU.

BibTeX

@article{jacobozavaleta2026thermal,
  author  = {Jácobo Zavaleta, Sergio and Copaci, Dorin and
             Die-Pancorbo, Sofía and Blanco, Dolores and Moreno, Luis},
  title   = {Indirect Thermal-Aware Supervision for
             Agonist--Antagonist SMA Ankle Actuation},
  journal = {Simposios del Comité Español de Automática (CEA)},
  volume  = {2},
  number  = {2},
  year    = {2026},
  month   = {5},
  doi     = {10.64117/simposioscea.v2i2.208}
}