Sensorimotor adaptation involves concurrent slow and fast processes, commonly modeled using a dual-state framework. This study employed a dual-state Kalman filter, incorporating both deterministic (interior point algorithm) and stochastic optimization methods, including genetic algorithm (GA), particle swarm optimization (PSO), and differential evolution algorithm (DEA). Five models were developed to estimate parameters and internal states (slow and fast) during a visual feedback treadmill walking task, using knee kinematics data from eleven participants. PSO achieved the highest accuracy in both full-state and internal-state estimations. GA showed the lowest full-state accuracy but performed well in internal-state estimation. DEA ranked second in full-state accuracy but struggled with internal states, while the interior point method had lower full-state accuracy but outperformed DEA in internal-state estimation. Notably, internal-state accuracy did not always align with full-state performance. Overall, stochastic methods outperformed the deterministic approach in estimating dual-state Kalman filter models for sensorimotor adaptation.

Data-driven methods are becoming popular to identify large, complex, and nonlinear systems and replacing equation-based methods. Dynamic mode decomposition methods are utilized to identify the linear dynamics of an unforced reference generator system and a highly nonlinear triple-link inverted pendulum biomechanical model around an equilibrium point. The identification methods successfully revealed the underlying linear dynamics of the systems. Feedforward and feedback optimal control policies are designed using a policy iteration method based on adaptive dynamic programming. The designed controllers exhibited satisfactory simulation performance in tracking the sit-to-stand movement of the nonlinear model.

As far back as Weiner, orthonormal basis functions were proposed for expressing the response of linear dynamical systems. Later, such functions were found useful for controller design in the case of linear time-invariant systems. This study explores the use of orthonormal functions in model predictive controller design for biomechanical sit-to-stand (STS) movement. Due to its simplicity and flexibility, model predictive control (MPC) has been popular in process control applications. MPC assumes finite prediction and control horizons, and optimizes a quadratic cost of state and incremental control variables subject to state and input constraints. In this study, MPC controller design technique is applied to a multi-segment biomechanical model, tasked to mimic physiological STS movement. Two sets of orthogonal functions, that is, Laguerre and Kautz functions, were explored. In both cases, MPC design generated physiologically constrained torques at ankle, knee, and hip joints that facilitated smooth STS movement. The MPC framework can be extended to controller design for other voluntary movements in a model based environment.

BACKGROUND: Dermoscopy is one of the common and effective imaging techniques in diagnosis of skin cancer, especially for pigmented lesions. Accurate skin lesion border detection is the key to extract important dermoscopic features of the skin lesion. In current clinical settings, border delineation is performed manually by dermatologists. Operator based assessments lead to intra- and inter-observer variations due to its subjective nature. Moreover it is a tedious process. Because of aforementioned hurdles, the automation of lesion boundary detection in dermoscopic images is necessary. In this study, we address this problem by developing a novel skin lesion border detection method with a robust edge indicator function, which is based on a meshless method. RESULT: Our results are compared with the other image segmentation methods. Our skin lesion border detection algorithm outperforms other state-of-the-art methods. Based on dermatologist drawn ground truth skin lesion borders, the results indicate that our method generates reasonable boundaries than other prominent methods having Dice score of 0.886 ±0.094 and Jaccard score of 0.807 ±0.133. CONCLUSION: We prove that smoothed particle hydrodynamic (SPH) kernels can be used as edge features in active contours segmentation and probability map can be employed to avoid the evolving contour from leaking into the object of interest.

Successful execution of biomechanical sit-to-stand (STS) task combines a forward thrust phase with an upward extension phase and stable movement termination. We have previously developed a fuzzy dynamic model to analyze the STS task by joining two local linear models, defined at the equilibrium positions, via Gaussian membership functions. The local linear models were obtained from a four-segment biomechanical representation of the human body dynamics in the sagittal plane. Our fuzzy controller model uses an observer to reconstruct velocity data from noisy observation of joint positions. In this study, we propose a reduced order observer with an optimal controller design for the STS task. The fuzzy optimal controller generates feedback and feedforward components of joint torques, whereby the latter are derived from a reference trajectory. Our movement control strategy employing fuzzy reduced order observer with fuzzy controller leads to physiologically tractable simulation of the STS movement with results that are superior to those previously obtained with full order compensators.

Human body walking movement involves both single and double support phases and is considered difficult to model. The aim of this study was to develop a method to capture human body dynamics during walking using Recurrent Neural Networks (RNN). In addition, a novel method using persistent excitation data generator is proposed to generate kinematic data to train the RNN in the absence of laboratory measurements. Kinematic data were applied to human body mathematical model to obtain required joint torques during bipedal walking. The RNN was used to approximate human body kinematics resulting from the joints torques for the walking movement. In order to test validity of the RNN model, model output was compared with human walking data captured in the laboratory. Simulation results show the model was able to approximate the joint angles during human walk with a low (10-4m) mean squared error for one stride.

Human biomechanical Sit-to-stand (STS) movement is a complex physiological task which requires movement coordination, balance and skillful termination. In previous studies we proposed the STS movement with different control strategies for physiologically relevant and optimized movements. These control techniques allowedus to optimize control systems designs for cost and robust performance. In this paper we used a four link biomechanical model with mixed H2/H∞control theory for STS transfer. We solved a biomechanical control problem with Linear Matrix Inequalities (LMI) for better physiological results by optimizing cost and bounds. Introducing LMIs for STS transfer provides a diverse perspective to physiological understand the movement. We compared these results with previous studies for H2and H∞control problems, and discussed the validity of LMI for human STS problem. Our simulation results shows that the biomechanical model with LMI control will provide a robust framework to analyze STS movement with physiological relevant control objectives.

Sit-to-stand (STS) movement is a common human task which involves combination of a musculoskeletal structure integrated with neural control. We present a conceptual 3D bipedal nonlinear model for STS task with optimal controller design for exoskeleton torques. This model has 7 sagittal plane angles, 3 frontal plane angles, and 3 foot translational variables with 3 position based holonomic constraints. We regulate the model with optimal controller design by coupled torque optimization due to muscular interaction between ankle, knee and hip joints. Our simulation results show the improvement in the results from previous controller design schemes. We further obtain experimental data for STS task on a force plate to compare the ground reaction forces of experimental data with our mathematical framework. This proves the validity of the model to extend this mathematical framework for further analysis of healthy and neuro-deficient subjects and accordingly designs of ergonomics and rehabilitation robotics.

Human biomechanical movements are complex physiological tasks which are efficiently regulated by the central nervous system (CNS). Proprioceptors (muscle spindles) provide feedback of fascicle length and velocity from a joint to CNS, which then control the entire movement. These feedbacks have delays which are accounted for by the required output command. In this study, we used a four-link sagittal plane nonlinear biomechanical model with three joint angles, to simulate human sit-to-stand (STS) movement in the presence of these physiological latencies. Ankle, knee and hip joint angles have delays for angular and velocity feedbacks. We linearized the whole model using padé approximation at sitting and standing positions which resulted in two eighteenth order linear systems. We integrated these local models into a fuzzy model with Gaussian membership function. The knee flexion angle during sit to stand movement provides the criterion for determining the weights of fuzzy membership functions. We developed a H2dynamic optimal controller for each local linear model and integrated with the fuzzy model. This controller computed the joint's torque or inputs for biomechanical STS task. We also introduced a reference trajectory to track the knee flexion angle error for smooth and physiological relevant sit to stand transfer. Simulation results of angular profiles and kinematics variables demonstrate the applicability of the fuzzy modeling with H2controller in the presence of feedback latencies.

This paper presents a wavelet-based recurrent fuzzy neural networks (WRFNN) trained with a stochastic search-based adaptation algorithm. A WRFNN represents a recurrent network of neurons employing wavelet functions whose outputs are combined using fuzzy rules. In this paper an earlier WRFNN model proposed by Lin, and Chin (2004), is modified by application of simultaneously perturbed stochastic approximation (SPSA) method for training the network. The model includes TSK-type fuzzy implication to compute output of each layer. The SPSA algorithm was shown to be a stable global optimization technique that is applicable to WRFNN models with demonstrated computational advantages over other optimization algorithms.

Many high-order systems have a large state space. Such systems need to additional computation time for complex calculation to find the output response. Traditionally, iteration methods have been applied to solve this problem. In this paper advantages of stability equation method derived by Parmer, [1], and the error minimization technique used in genetic-fuzzy algorithm have been combined to propose a new method for order reduction of linear dynamic systems described via state-space models. Genetic part has been used in this formulation to find the optimal solution(s) to minimize the objective function ¿J¿ that depends on the error term between the original output and the desired or reduced output. Fuzzy sets have been used to determine the step size action (point crossover or multiple crossover) depending upon fuzzy rules based on the current and previous error terms. An example of reduced order modeling from power systems is presented to illustrate the algorithm.

Augmented reality (AR) system is a virtual environment that overlays the computer generated graphics on real-world view. A typical AR system consists of optical motion tracking and capturing. The main problem in such a system is minimizing error of mapping the movements of the actor in real domain to the virtual model. To do that, the position of markers attached to the actor needs to be carefully selected and mapped to the virtual counterpart. This one-to-one mapping process is called registration. In this regard, virtual and real domains should be aligned correctly. Any minute error in registration of these two domains results in degrading of the realism e.g. wrongly registered objects floating through each other. Current registration method for optical motion capture systems requires time-consuming and tedious manual processing. To overcome this problem, we present two-way (virtual object to real domain and real domain to virtual object) semi-automatic registration method. With our method instead of priory determining positions of markers on the physical object, we determine markers' positions (extreme points found by our application) on the virtual object and then locate markers at the approximate positions on the real object. By tracking the markers in the following step, we get markers' positions in real domain and change virtual domain's markers' positions accordingly to reduce error. Results show that our system solves the registration problem and prevents unrealistic jittering and flickering effects due to misalignment.

The neuro-physiological mechanisms involved in postural stabilization are not well understood. Human body mechanically resembles an inverted pendulum that is inherently unstable. Active and passive mechanisms at muscle and spinal level as well as visual and vestibular processes are attributed to postural stabilization. At the same time, intrinsic delays in the reflex pathways and the low-pass characteristics of the muscle response tend to limit the effectiveness of active mechanisms of stabilization. The motivation for this research was to study the relative contribution from active control (feedforward mechanisms) and passive stiffness (feedback mechanisms) in maintenance of posture and coordination of voluntary movement. We develop a multi-segment sagittal model with three degrees of freedom that included the rotation at ankle, knee, and hip joints. We propose an optimal LQR controller as the central nervous system analog in posture and movement coordination. We present analytical and simulation results to support an active-passive model of postural stabilization. Besides expanding our understanding of the stabilization processes in the body, the insight gained from this study is expected to promote awareness of the existence of optimal trajectories in the coordination of skilled voluntary movements.

A process of determining whether two or more bodies are making contact at one or more points is called collision detection or intersection detection. Collision detection is inseparable part of the computer graphics, surgical simulations, and robotics. There are varieties of methods for collision detection. We will review some of the most common ones. Algorithms for contact determination can be grouped into two general parts: broad-phase and narrow-phase. This paper provides a comprehensive classification of a collision detection literature into the two phases. Moreover, we have attempted to explain some of the existing algorithms which are not easy to interpret. Also, we have tried to keep sections self-explanatory without sacrificing depth of coverage.

The neuro-physiological mechanisms involved in postural stabilization are not well understood. Human body mechanically resembles an inverted pendulum that is inherently unstable. Active and passive mechanisms at muscle level, as well as other visual and vestibular processes, are attributed to stability. The available evidence suggests that muscle stiffness alone is insufficient to stabilize body sway, and must rely on active mechanisms of stabilization that are unlikely to have a reflex nature due to the intrinsic delays in the reflex pathways and the low-pass characteristics of the muscle response. The role played by the central nervous system in active control of stance thus remains an open and intriguing question. In this study we present simulation results to support an active-passive model of postural stabilization. Besides expanding our understanding of the postural stabilization process, the insight gained would be useful to promote intervention techniques for therapists and clinicians working with fall-prone individuals.

In this paper we address the problem of PID controller stabilization of a single-link inverted pendulum-based biomechanical model with force feedback, two levels of position and velocity feedback, and with delays in all the feedback loops. The motivation for this work arises from considering the postural stabilization problem in biomechanics where the human body is modeled as a neuro-musculo-skeletal system. The proprioceptive feedback from muscle spindle and Golgi tendon organ generates short, medium, and long latency responses from the central nervous system (CNS), consisting of the brain and the spinal chord. Of these responses the first two are included in the formulation. The Hermite-Biehler Theorem is used to derive stability results, leading to necessary and sufficient conditions for existence of stabilizing PID controllers for the model. An algorithm for selection of stabilizing gains is developed.

The coordination problem of a human arm and a robot manipulator is explored using compliant motion and predictive control. The problem arises when a human arm and a robot manipulator coordinate for the execution of a task in unscheduled task environment. In such scenarios the arm, by virtue of its intelligence, is assumed to lead the task while the manipulator is required to comply with the motion of the arm and support the object load. Such a scheme is superior to the better known multiple manipulator coordination problem, which normally assumes known trajectories and a structured task environment. By coordinating manipulator with the arm of its operator the uncertainty due to the environment can be reduced while load sharing can help relieve the arm of the physical strain. This paper addresses the problem in the framework of model-based predictive control. The transfer function from the manipulator position command to the wrist sensor force output is defined. The desired set point for the manipulator force is set to equal the gravitational force. A predictive control scheme is then used to design a two-degree of freedom controller for the problem. The simulation results indicate that the manipulator effectively takes over the object load and the arm force stays close to zero. Moreover, the manipulator is seen to be highly responsive to the arm movement and relatively small arm force can effectively initiate the manipulation task.

Simultaneous stabilization and disturbance decoupling of constrained robotic systems with minimal inputs are addressed. The robotic system must be decoupled into disjoint manifolds of motion and constraint, and the small motion of the system in the respective tangential hyperplanes must be stable. The problem arises in rolling and gliding types of probing manipulation. Necessary conditions for block decoupling with minimum inputs require that the image of the input matrix have a nonnull component along the null space of a matrix function of the constraint equations. Necessary conditions for simultaneous stabilization and decoupling with minimal inputs are established. An example of a planar four-link biped robot and a method to avoid constraint surface penetration are presented.>