Chris Wilson Antuvan

Adaptive backlash compensation in upper limb soft wearable exoskeletons

A new frontier of assistive devices aims at designing exoskeletons based on fabric and flexible materials for applications where kinematic transparency is the primary requirement. Bowden-cable transmission is the widely employed solution in most of the aforementioned applications due to advantages in durability, lightweight, safety, and flexibility. The major advantages of soft assistive devices driven by bowden-cable transmissions can be identified in the superior ergonomics and wearability, allowing users to freely move and allocating the actuation stages far from the end-effector. However, control accuracy in bowden-cable transmission presents some intrinsic limitation due to nonlinearities such as static and dynamic friction, occurring between the cables and the bowden sheaths, and backlash hysteresis. Friction and backlash effects are known to be related to the curvature of the flexible sheath, which is not directly measurable and can vary during human motion. In this paper we describe our new wearable exosuit for upper limb assistance and in particular we introduce a mathematical model for backlash hysteresis compensation. The implementation of a nonlinear adaptive controller is described in detail and experimentally tested on the proposed design as a backlash compensation strategy: results report that the adaptive controller improves the accuracy in position tracking (i.e.RMSE in trajectory tracking 1) by compensating for time-varying backlash and continuously updating the model parameters. The backlash hysteresis model and the proposed control scheme are validated first on a custom-designed test bench and then applied to control the soft exoskeleton worn by a subject affected by bilateral brachial plexus injury. Soft exosuit driven by Bowden cables for human arm assistance.Backlash hysteresis in Bowden-cable transmission.Nonlinear adaptive controller for backlash compensation.

Role of Muscle Synergies in Real-Time Classification of Upper Limb Motions using Extreme Learning Machines

BackgroundMyoelectric signals offer significant insights in interpreting the motion intention and extent of effort involved in performing a movement, with application in prostheses, orthosis and exoskeletons. Feature extraction plays a vital role, and follows two approaches: EMG and synergy features. More recently, muscle synergy based features are being increasingly explored, since it simplifies dimensionality of control, and are considered to be more robust to signal variations. Another important aspect in a myoelectrically controlled devices is the learning capability and speed of performance for online decoding. Extreme learning machine (ELM) is a relatively new neural-network based learning algorithm: its performance hasn’t been explored in the context of online control, which is a more reliable measure compared to offline analysis. To this purpose we aim at focusing our investigation on a myoelectric-based interface which is able to identify and online classify, upper limb motions involving shoulder and elbow. The main objective is to compare the performance of the decoder trained using ELM, for two different features: EMG and synergy features.MethodsThe experiments are broadly divided in two phases training/calibration and testing respectively. ELM is used to train the decoder using data acquired during the calibration phase. The performance of the decoder is then tested in online motion control by using a simulated graphical user interface replicating the human limb: subjects are requested to control a virtual arm by using their muscular activity. The decoder performance is quantified using ad-hoc metrics based on the following indicators: motion selection time, motion completion time, and classification accuracy.ResultsPerformance has been evaluated for both offline and online contexts. The offline classification results indicated better performance in the case of EMG features, whereas a better classification accuracy for synergy feature was observed for online decoding. Also the other indicators as motion selection time and motion completion time, showed better trend in the case of synergy than time-domain features.ConclusionThis work demonstrates better robustness of online decoding of upper-limb motions and motor intentions when using synergy feature. Furthermore, we have quantified the performance of the decoder trained using ELM for online control, providing a potential and viable option for real-time myoelectric control in assistive technology.

Hierarchical Cascade Controller for Assistance Modulation in a Soft Wearable Arm Exoskeleton

In recent years soft wearable exoskeletons, commonly referred to as exosuits, have been widely exploited in human assistance. Hence, a shared approach for a systematic and exhaustive control architecture is extremely important. Most of the exosuits developed so far employ a bowden cable transmission to conveniently place the actuator away from the end-effector. While having many advantages this actuation strategy presents some intrinsic limitations caused by the presence of nonlinearities, such as friction and backlash of the cables, which make it difficult to predict and control the dynamics between the device and the user. In this letter, we propose a novel hierarchical control paradigm for a cable-driven upper limb exosuits that aims at evaluating and consequently deliver the appropriate assistive torque to the users elbow joint. The proposed control method comprises three main layers: an active impedance control which estimates the users arm motion intention and guarantees an intuitive response of the suit to the wearers motion; a mid-level controller which compensates for the backlash in the transmission and converts the reference arm motion to the desired position of the actuator; a low-level controller which is responsible for driving the actuation stage by compensating for the nonlinear dynamics occurring in the bowden cable to provide the desired assistive torque at the joint. Tests on healthy subjects show that wearing the exosuit reduces by

Preliminary design and control of a soft exosuit for assisting elbow movements and hand grasping in activities of daily living

48.3\%

Design and Preliminary Testing of a Soft Exosuit for Assisting Elbow Movements and Hand Grasping

the muscular effort required to lift 1 kg and that the controller is able to modulate its level of assistance to the wearers motor ability.

EMG-based learning approach for estimating wrist motion.

The development of a portable assistive device to aid patients affected by neuromuscular disorders has been the ultimate goal of assistive robots since the late 1960s. Despite significant advances in recent decades, traditional rigid exoskeletons are constrained by limited portability, safety, ergonomics, autonomy and, most of all, cost. In this study, we present the design and control of a soft, textile-based exosuit for assisting elbow flexion/extension and hand open/close. We describe a model-based design, characterisation and testing of two independent actuator modules for the elbow and hand, respectively. Both actuators drive a set of artificial tendons, routed through the exosuit along specific load paths, that apply torques to the human joints by means of anchor points. Key features in our design are under-actuation and the use of electromagnetic clutches to unload the motors during static posture. These two aspects, along with the use of 3D printed components and off-the-shelf fabric materials, contribute to cut down the power requirements, mass and overall cost of the system, making it a more likely candidate for daily use and enlarging its target population. Low-level control is accomplished by a computationally efficient machine learning algorithm that derives the system’s model from sensory data, ensuring high tracking accuracy despite the uncertainties deriving from its soft architecture. The resulting system is a low-profile, low-cost and wearable exosuit designed to intuitively assist the wearer in activities of daily living.

Muscle synergies for reliable classification of arm motions using myoelectric interface

Most of the currently available exoskeletons for upper limbs are constrained by limited portability, ergonomics, weight and, energy-wise, autonomy. Moreover, their high cost makes them available only for the most affluent users, ruling out the majority of the population in need. By replacing rigid aluminum links and transmissions with fabrics and bowden cables, we can both cut down the cost of the assistive device and design it to be portable, comfortable and lightweight. We present the design and a preliminary testing of a soft exosuit for assisting elbow flexion/extension and hand open/close. Our system comprises two proximally located tendon-driving actuators, and two textile-based frames that route the tendons and transmit forces to the human joints, namely an elbow sleeve and a glove. A preliminary test on a healthy subject is presented with an adaptive controller that achieves good tracking accuracy despite of the system’s non-linear and time-varying dynamics.

The Influence of External Forces on Wrist Proprioception

This paper proposes an EMG based learning approach for estimating the displacement along the 2-axes (abduction/adduction and flexion/extension) of the human wrist in real-time. The algorithm extracts features from the EMG electrodes on the upper and forearm and uses Support Vector Regression to estimate the intended displacement of the wrist. Using data recorded with the arm outstretched in various locations in space, we train the algorithm so as to allow robust prediction even when the subject moves his/her arm across several positions in space. The proposed approach was tested on five healthy subjects and showed that a R2 index of 63.6% is obtained for generalization across different arm positions and wrist joint angles.

Simultaneous classification of hand and wrist motions using myoelectric interface: Beyond subject specificity

Synergistic activation of muscles are considered to be the phenomenon by which the central nervous system simplifies its control strategy. Muscle synergies are neurally encoded and considered robust to be able to adapt for various external dynamics. This paper presents a myoelectric-based interface to identify and classify motions of the upper arm involving the shoulder and elbow. We contrast performance of the decoder while using time domain and synergy features. The decoder is trained using extreme learning machine algorithm, and online testing is performed in a virtual environment. Better classification accuracy for online control is obtained while using muscle synergy features. The results indicate better online performance compared to offline performance while using synergy features to classify movements, indicating generalization to dynamic situations and robustness of control.

Discrete classification of upper limb motions using myoelectric interface

Proprioception combines information from cutaneous, joint, tendon, and muscle receptors for maintaining a reliable internal body image. However, it is still a matter of debate, in both neurophysiology and psychology, to what extent such body image is modified or distorted by a changing haptic environment. In particular, what is worth investigating is the contribution of external forces on our perception of body and joint configuration. The proprioceptive acuity of fifteen young participants was tested with a Joint Position Matching (JPM) task, performed with the dominant wrist under five different external forces, in order to understand to what extent they affect proprioceptive acuity. Results show that accuracy and precision in target matching do not change in a significant manner as a function of the loading condition, suggesting that the multi-sensory integration process is indeed capable of discriminating different sub-modalities of proprioception, namely the joint position sense and the sense of force. Furthermore, results indicate a preference for target undershooting when movements are performed in a viscous or high resistive force field, rather than passive or null fields in which subjects did not show any predominance for under/over estimation of their position.