Design and Implementation of a Position, Speed and Orientation Fuzzy Controller Using a Motion Capture System to Operate a Wheelchair Prototype

Abstract

The design and implementation of an electronic system that involves head movements to operate a prototype that can simulate future movements of a wheelchair was developed here. The controller design collects head-movements data through a MEMS sensor-based motion capture system. The research was divided into four stages: First, the instrumentation of the system using hardware and software; second, the mathematical modeling using the theory of dynamic systems; third, the automatic control of position, speed, and orientation with constant and variable speed; finally, system verification using both an electronic controller test protocol and user experience. The system involved a graphical interface for the user to interact with it by executing all the controllers in real time. Through the System Usability Scale (SUS), a score of 78 out of 100 points was obtained from the qualification of 10 users who validated the system, giving a connotation of “very good”. Users accepted the system with the recommendation to improve safety by using laser sensors instead of ultrasonic range modules to enhance obstacle detection.

Using Inertial Sensors to Determine Head Motion—A Review

“…In DLMs, this architecture enables the recognition of more complex activities, and it does not require an additional preprocessing step. In CML models, the architectures used most in the field of head motion detection are regression models (RMs) [34], random forest (RF) [36], feedforward artificial neural networks (FANNs) [58,63,75], dynamic time warping (DTW) [76], decision tree (DTs) [28,36], support vector machines (SVMs) [42,64], k-nearest neighbor (k-NN) [46], fuzzy logic (FL) [79], naïve Bayes classifier (NBC) [50,51,62], Euclidian distance classifiers (EDCs) [54], Mahalanobis distance classifiers (MDCs) [54], Gaussian mixture models (GMs) [25], Gauss-Newton models (GNMs) [49], adaptive boosting classifiers (AD-ABs) [80], and multilayer perceptron (MLP) classifiers [81]. As for DLM models, the most common deep learning models are long short-term memory networks (LSTMs), convolutional neural networks-long short-term memory networks (CNNs-LSTMs), convolutional neural networks (CNNs), bidirectional LSTM networks (BLSTMs), convolutional neural networks-bidirectional LSTM networks (CNNs-BLSTMs) [29,57], and hidden Markov models [55].…”

“…After the literature review, we noticed that the relevant selected papers had the following distribution: 49% focused on medical problems such as head tremor [34], cerebral palsy [30][31][32][33], fall detection [15,34], vestibular rehabilitation [35], physical and mental activity analysis [20,36,37], forward head posture [17,38], and musculoskeletal disorders [39]; 20% focused on the general problem of human-computer interaction [35,[40][41][42]; and 10% focused on the development of new computational and calibration methods [39,43,44]. The last two topics were the development of sports devices (swimming, hokey, golf, motorcycle ride or spinning exercises: 10%) [45][46][47][48][49][50], and 8% focused on prevention and safety systems (drivers' attention) [51][52][53]. All the selected papers used an electronic device with an inertial sensor placed on a specific part of the head.…”

“…On the other hand, in the unsupervised class, the focus is on the identification of head motion patterns from the input dataset without any knowledge of the desired output. Based on the papers related to this study, the most common classical machine learning algorithms are regression models (RMs) [34], random forest (RF) [36], feedforward artificial neural network (FANNs) [58,63,75], dynamic time warping (DTW) [76], decision tree (DTs) [28,36], support vector machines (SVMa) [42,64], knearest neighbor (k-NN) [46], fuzzy logic (FL) [79], naïve Bayes classifiers (NBCs) [50,51,62], Euclidian distance classifiers (EDCs) [54], Mahalanobis distance classifiers (MDCs) [54], Gaussian mixture (GM) models [25], Gauss-Newton models (GNMs) [49], adaptive boosting classifiers (ADABs) [80], and multilayer perceptron (MLP) classifiers [81]. Classical machine learning models are usually preferred in the field of head motion recognition or human activity recognition, especially when the training dataset is small, or when a rapid training process is necessary.…”

A Kinematic Information Acquisition Model That Uses Digital Signals from an Inertial and Magnetic Motion Capture System

“…For this reason, starting from the fact that inertial systems provide information in technical language, this article proposes a method that allows the signals generated by an inertial-magnetic motion capture system to be converted into kinematic information (expressed in degrees corresponding to the specific movements of the body segment) that can be easily interpreted by professionals in the field of physical rehabilitation. To achieve this, the authors use a system named Imocap-GIS [27][28][29], which has been used in several projects and has made a significant contribution to the use of inertial-magnetic motion capture systems [30][31][32][33][34]. The use of this technology allows the use of a method to obtain kinematic information directly, reducing the required time, and lessening the data analysis and interpretation processes since specialized personnel and expertise are not always required.…”

Design and Optimization of a Neuro-Fuzzy System for the Control of an Electromechanical Plant

“…The literature reviewed shows that fuzzy systems are widely employed in multi-motor control. Meanwhile, in [11] a fuzzy controller is proposed using a motion capture system to operate the speed and orientation of a wheelchair considering the model of a Direct Current (DC) motor for the design process. Further, a fuzzy logic controller is used in [12], for an image-based navigation system of an autonomous underwater vehicle with a DC motor.…”