• actuator
• arm
• control
• elbow
• impedance
• prosthetic
• stiffness
• variable

The thesis contributes to the field of rehabilitation robotics, aiming to restore movement and functionality for individuals who have experienced limb loss. One prominent challenge involves achieving prosthetic behavior that is both natural in movement and interactions with people and objects. A key research question revolves around developing realistic decoding techniques for position and stiffness.
Observing human nature, there is a tendency towards adaptive impedance behavior in prosthetics. Consequently, the use of variable impedance actuators, capable of adjusting impedance based on the intended action, has gained prominence. However, the adoption of variable stiffness prosthetics introduces a significant computational burden, prompting a legitimate inquiry into the cost-effectiveness of choosing variable stiffness over fixed stiffness actuators.
This thesis delves into the investigation of this dual challenge, conducting experiments in tele-operation with electromyographic signals from the biceps and triceps of healthy subjects.
In the field of prosthetics, soft robotics emerges as an essential element, offering innovative solutions and significant benefits to improve effectiveness and user experience. These include adaptability and flexibility that allow natural movement and improved control of interaction with the external environment. The difference between a common rigid robot, where the link is directly connected to the gearbox, and a robot with VIA (Variable Impedence Actuator), is that the link is decoupled from the motor inertia by an adjustable spring/damper unit. VSA (Variable Stiffness Actuator) is a simplified version of a VIA with an adjustable spring unit, but without a mechanical damping unit. Adjusting the stiffness through a mechanical reconfiguration and controlling the position of one-DoF joint requires two motors.
The hardware used is the Vs-Elbow D, a single-degree-of-freedom VSA in an independent configuration: one motor is used to drive position output and the other for stiffness control. The Vs-Elbow D is composed of two segments: Upperside segment (U-Segment) and Forearm-side segment (F-Segment). Each segment includes a motor unit and its transmission mechanism. The F-segment is composed of the motor M1 and can rotate by attracting the Elbow Shaft . To rotate Segment F, M1 generates an action on the Elbow Shaft and when this action is compensated by the total action of the elastic element the shaft is fixed and Segment F can move around it. Segment U consists of the stiffness motor and the elastic transmisison. Two contributions are considered in calculating the forearm joint angle: relative position, which gives the position of the joint shaft relative to the forearm, and joint deflection, which gives the position of the joint shaft relative to the upper-arm.
To operate an externally-powered prosthesis, the user directly commands the device by utilizing biosignals obtained from their body. The monitoring technique employed is electromyography (EMG), aimed at evaluating muscle functionality by analyzing the electrical potentials generated during muscle activity. In traditional systems, the primary method for managing a single-degree-of-freedom prosthesis is through proportional position control. This approach demands ongoing muscle activation to sustain the prosthesis in any position beyond the one associated with null EMG signals. To address this limitation, proportional velocity controls come into consideration. These controls define the prosthetic command relative to the integral of the EMG signal difference.
When manipulating task-related gripping forces using an impedance controller, adjusting co-contraction levels influences compliance. Lower co-contraction leads to increased compliance, facilitating gentle grasping of delicate or deformable objects. Conversely, higher stiffness values are applied during the grip of heavier and stiffer objects. This property enables the gradual adjustment of the estimated hand force. One approach to control stiffness involves determining the stiffness command in proportion to the overall activation of the user's muscles. This is calculated as the weighted average of the activation levels from two antagonistic muscles. Another employed method considers stiffness as the minimum value derived from the two electromyographic signals.
The experiments take up the methods of the work conducted by Blank, comparing the decoding techniques covered by Capsi-Morales.
Conducting the experiments has a twofold objective: first, to validate the position and stiffness control techniques, and second, to investigate the validity of variable impedance control. Each subject controls the prosthesis by surface EMG electrodes placed on the biceps and triceps. Through contractions and co-contractions, the user adjusts the position and stiffness of the Vs-Elbow through each of the 5 control strategies.
Each experiment follows the procedure below: Electrode placement: electrode placement heavily affects the quality of the electromyographic measurement and selectivity regarding the muscle origin of the EMG signal; EMG processing and calibration of thresholds: since the characteristics of muscle activation and maximum contraction force vary from subject to subject, EMG signals must be processed appropriately. Once the signal between zero and 1 is obtained, the calibration of the necessary thresholds for the controllers is carried out. The calibration procedure follows what Sensinger reported; Precision task: the goal is to reach a given position and maintain it following a perturbation that occurs at a random time;
The results obtained from these experiments suggest the efficacy of adjusting stiffness according to the specific task at hand.