Maria M. Martins

Hybridization between multi-objective genetic algorithm and support vector machine for feature selection in walker-assisted gait

Walker devices are often prescribed incorrectly to patients, leading to the increase of dissatisfaction and occurrence of several problems, such as, discomfort and pain. Thus, it is necessary to objectively evaluate the effects that assisted gait can have on the gait patterns of walker users, comparatively to a non-assisted gait. A gait analysis, focusing on spatiotemporal and kinematics parameters, will be issued for this purpose. However, gait analysis yields redundant information that often is difficult to interpret. This study addresses the problem of selecting the most relevant gait features required to differentiate between assisted and non-assisted gait. For that purpose, it is presented an efficient approach that combines evolutionary techniques, based on genetic algorithms, and support vector machine algorithms, to discriminate differences between assisted and non-assisted gait with a walker with forearm supports. For comparison purposes, other classification algorithms are verified. Results with healthy subjects show that the main differences are characterized by balance and joints excursion in the sagittal plane. These results, confirmed by clinical evidence, allow concluding that this technique is an efficient feature selection approach.

Estimating and controlling UAV position using RGB-D/IMU data fusion with decentralized information/Kalman filter

This paper proposes a 3D data capture system, based on the fusion of data coming from an active depth sensor and a inertial measurement unit (IMU), to determine the position of an aerial unmanned vehicle (UAV) in indoor environments, for control purposes. Firstly, the method adopted to detect the vehicle through using a sequence of RGB-D images. After that, the information provided by the active depth sensor is fused with the data provided by the IMU onboard the vehicle, using a Decentralized Kalman Filter (DKF) and a Decentralized Information Filter (DIF), whose performance are compared. In the sequel, a nonlinear controller is used for positioning the UAV. Finally, the performance differences between the DKF and the DIF are highlighted, as well as the divergence between the results of the depth sensor and the inertial one, in experiments involving abrupt maneuvers to induce estimation errors in the inertial unit, to check the effectiveness of the developed 3D data capture system.

Fast embedded feet pose estimation based on a depth camera for smart walker

Assisted gait monitoring could benefit from the measurement of feet position and orientation. These measurements could also be used to implement intention-based control laws for smart walkers. Several works have been dedicated to the detection of lower limbs. The proposed methods are usually fast, but only detect the position of the legs. Others may even be more complete, but are not adapted to reactive applications. Also, they often use markers attached on the feet, which is unsuitable in daily routine. In this paper, a fast feet position and orientation detection algorithm is proposed. It is based on a camera depth sensor and does not require the use of any marker. The obtained results are compared with a ground truth provided by a motion traking system to experimentally assess the performances of the proposed algorithm.

Legs tracking for walker-rehabilitation purposes

Clinical evaluation during walker-assisted gait is the first step to assess the evolution of a patient during rehabilitation and to identify his needs and difficulties. Advances in robotics made it possible to integrate a gait analysis tool on a walker to enrich the existing rehabilitation tests with new sets of objective gait parameters. This paper focuses on the legs detection method to estimate legs position during an assisted walk and the detection of gait events. In this paper, a walker is equipped with a laser range sensor (LRF) and encoders to analyze the spatiotemporal parameters of the walker users. Preliminary results obtained on ten subjects show that relevant data using a LRF can be extracted for gait analysis with a small error.

Dynamical system approach for obstacle avoidance in a Smart Walker device

The need for Smart Walkers to help navigation of elderly is increasing rapidly. This paper proposes an approach based on the Dynamical System Approach for obstacle avoidance of a Smart Walker device. Layers of sonars are distributed on the walker for detecting obstacles and stairs. A simulated model of the ASBGo Walker [1] and several realistic simulations in a hospital environment with typical hospital obstacles illustrate the good performance of the approach.

Feature reduction with PCA/KPCA for gait classification with different assistive devices

Purpose – The purpose of this paper is to propose a gait analysis technique that aims to identify differences and similarities in gait performance between three different assistive devices (ADs). Design/methodology/approach – Two feature reduction techniques, linear principal component analysis (PCA) and nonlinear kernel-PCA (KPCA), are expanded to provide a comparison of the spatio-temporal, symmetrical indexes and postural control parameters among the three different ADs. Then, a multiclass support vector machine (MSVM) with different approaches is designed to evaluate the potential of PCA and KPCA to extract relevant gait features that can differentiate between ADs. Findings – Results demonstrated that symmetrical indexes and postural control parameters are better suited to provide useful information about the different gait patterns that total knee arthroplasty (TKA) patients present when walking with different ADs. The combination of KPCA and MSVM with discriminant functions (MSVM DF) resulted in a n...

A novel human-machine interface for guiding: The NeoASAS smart walker

In an aging society it is extremely important to develop devices, which can support and aid the elderly in their daily life. This demands tools that extend independent living and promote improved health. In this work it is proposed a new interface approach integrated into a walker. This interface is based on a force sensor and it is intended to extract the users movement intentions. The interface is designed to be user-friendly, simple and intuitive, efficient and economic, meeting usability aspects and focused on a commercial implementation, but not being demanding at the user cognitive level. Preliminary sets of experiments were performed which showed the sensibility of the force sensor extract navigation commands from the user. These signals presented a higher frequency component that was attenuated by a Benedict-Bordner g-h filter. The presented methodology offers an effective cancelation of the undesired components from force sensor data, allowing the system to extract in real-time voluntary users navigation commands. Based on this real-time identification of voluntary users commands, an approach to the control architecture of the robotic walker is being developed, in order to obtain stable and safe user assisted locomotion.

Development of a Biofeedback Approach Using Body Tracking with Active Depth Sensor in ASBGo Smart Walker

There is a considerable number of people with balance problems. Thus, the development of devices that can support and aid patients these patients is a very important question in our days. One important point to take into account during their treatment is which relevant information should be presented to them to improve their motivation and movement control. Biofeedback is an important tool since it shows to the patient his/her real performance [1]. This article tackles the challenge of integrating such a tool in a smart walker. It presents an application that was developed to aid patients with balance problems while walking assisted with the Adaptive System Behavior Group (ASBGo) smart walker (SW). Through a built-in Active Depth Sensor (ADS), it is possible to recognize the upper body and extract data about the users performance. Such information is processed with the developed application and displayed in a graphics interface in real time. Results show the high potential of this application to be used during a clinical trial using the smart walker as a rehabilitation tool.

Mechatronic medical device for wrist rehabilitation

In the paper, the mechatronic design of a new type of Powerball®, named ControlledBioBall, carried out to overcome the present drawbacks, is be presented and discussed with special focus on the software developed in the editor LabVIEW®. Devices for wrist rehabilitation must be able to make movements of the wrist quirky and complex. Powerball® is such a market existing device that can be used for wrist rehabilitation and also as a sports aid for strengthening the wrist. The operation of Powerball® is based on the gyroscope principle. While the gyroscope is spinning, the movement of the wrist will tilt the shell and a reaction moment will be produced. Nevertheless, Powerball® in the existing market have considerable limitations, such as: must be manually started; the rotation speed is not controlled; the torque must be at least equal to the minimum torque to maintain the rotation of the mass inside Powerball®, and finally, due to the small panel attached to the ball patients will find inconvenience to view and manipulate it. This manner, for improving the effectivity of the wrist rehabilitation processes, a new type of Powerball® has been developed. This will have adjustable speed given by an electric motor inside the ball, and consequently the manual start and the requirement of minimum torque will be eliminated. Also, a system composed by sensors, data acquisition board and human-machine interface will be considered to allow the process of rehabilitation to be fully controlled..

Evaluation of gait performance of knee osteoarthritis patients after total knee arthroplasty with different assistive devices

IntroductionNowadays Knee Osteoarthritis (KOA) affects a large percentage of the elderly, and one solution is to perform a Total Knee Arthroplasty (TKA). In this paper, one intends to study the gait and posture of these patients after the TKA, while walking with three assistive devices (ADs) (crutches, standard walker (SW) and rollator with forearm supports (RFS)).MethodsEleven patients were evaluated in 2 phases: 5 days and 15 days after surgery. This evaluation was conducted with two inertial sensors, one attached to the operated leg ankle, to measure spatiotemporal parameters, and the other at the sacrum, to measure posture and fall risk-related parameters. Multivariate analysis of variance (MANOVA) with repeated measures was performed to detect group differences.ResultsThe MANOVA results show that all spatiotemporal parameters are significantly different (p 0.05). The interaction between time and ADs only affects significantly the velocity (p<0.05). In terms of fall risk parameters, time only significantly affects the antero-posterior direction (p<0.05) and ADs affects significantly root mean square in medio-lateral direction (p<0.05). In terms of interaction between time and ADs, there are no statistical significant differences.ConclusionThis study concludes that depending on the state of recovery of the patient, different ADs should be prescribed. On the overall, standard walker is good to give stability to the patient and RFS allows the patient to present a gait pattern closer to a natural gait.