Browsing by Author "Trythall, Simon"

Now showing 1 - 3 of 3
  • Thumbnail Image
    ItemOpen Access
    A two-stage classification method for improved positioning using low-cost inertial sensors
    (IEEE, 2024-08-08) Maton, Dariusz; Economou, John; Galvao Wall, David; Khan, Irfan; Cooper, Robert; Ward, David; Trythall, Simon
    The two-stage subtractive clustering Takagi-Sugeno (2SC-TS) method is proposed which completely replaces the established method of inertial navigation with classification models. The classifiers are designed by the subtractive clustering algorithm, an unsupervised learning method. The accuracy of the trajectories is compared against three competitive data-driven methods on three independent experimental datasets. The results show how 2SC-TS generates trajectories with approximately 20% lower positional error compared with the single-stage version (SC-TS), and halves the error produced by competitive deep learning methods. The proposed method may help improve the positioning of people and robots carrying low-cost inertial sensors.
  • Thumbnail Image
    ItemOpen Access
    Indirect tuning of a complementary orientation filter using velocity data and a genetic algorithm
    (Taylor and Francis, 2024-04-23) Maton, Dariusz; Economou, John T.; Galvão Wall, David; Khan, Irfan; Cooper, Robert; Ward, David; Trythall, Simon
    In this paper, the accuracy of inertial sensor orientation relative to the level frame is improved through optimal tuning of a complementary filter by a genetic algorithm. While constant filter gains have been used elsewhere, these may introduce errors under dynamic motions when gyroscopes should be trusted more than accelerometers. Optimal gains are prescribed by a Mamdani fuzzy rule base whose membership functions are found using a genetic algorithm and experimental data. Furthermore, model fitness is not based directly on orientation but the error between estimated and ground truth velocities. This paper has three interrelated novel elements. The main novelty is the indirect tuning method, which is simple, low-cost and requires a single camera and inertial sensor. The method is shown to increase tracking accuracy compared with popular baseline filters. Secondary novel elements are the bespoke genetic algorithm and the time agnostic velocity error metric. The contributions from this work can help improve the localization accuracy of assets and human personnel. This research has a direct impact in command and control by improving situational awareness and the ability to direct assets to safe locations using safer routes. This results in increasing safety in applications such as firefighting and battlespace.
  • Thumbnail Image
    ItemOpen Access
    Subtractive clustering Takagi-Sugeno position tracking for humans by low-cost inertial sensors and velocity classification
    (Sage, 2023-04-12) Maton, Dariusz; Economou, John T.; Galvão Wall, David; Ward, David; Trythall, Simon
    In this work, open-loop position tracking using low-cost inertial measurement units is aided by Takagi-Sugeno velocity classification using the subtractive clustering algorithm to help generate the fuzzy rule base. Using the grid search approach, a suitable window of classified velocity vectors was obtained and then integrated to generate trajectory segments. Using publicly available experimental data, the reconstruction accuracy of the method is compared against four competitive pedestrian tracking algorithms. The comparison on selected test data, has demonstrated more competitive relative and absolute trajectory error metrics. The proposed method in this paper is also verified on an independent experimental data set. Unlike the methods which use deep learning, the proposed method has shown to be transparent (fuzzy rule base). Lastly, a sensitivity analysis of the velocity classification models to perturbations from the training orientation at test time is investigated, to guide developers of such data-driven algorithms on the granularity required in an ensemble modelling approach. The accuracy and transparency of the approach may positively influence applications requiring low-cost inertial position tracking such as augmented reality headsets for emergency responders.