Sooraksa, Pitikhate

Preferred name

Sooraksa, Pitikhate

Alternative Name

Sooraksa, P.

Main Affiliation

School of Engineering

pitikhate.so@kmitl.ac.th

Now showing 1 - 10 of 26

Some of the metrics are blocked by your
consent settings
Physics-informed Platform for Flight Dynamics Simulation
(2025-01-01)
Atayagul, Pattiwat
;
Sooraksa, Pitikhate
In this study, we present the state-of-the-art development of the flight dynamics model of a rigid body from the theoretical aspect of flight dynamics, design the software architecture, and validate the proposed architecture by comparing the simulation results with the check-cases for the verification of six-degree-of-freedom flight vehicle simulations document issued by the National Aeronautics and Space Administration (NASA). We also design the computational workflow between the ordinary differential equations of the system and other axillary components by weaving those relations in the object-oriented programing style and powered with Python scientific libraries. The simulation outcomes in all cases are well matched with the majority of NASA baseline datasets under the same flight simulation conditions, which reflect the accuracy of the model with considerable confidence.
Some of the metrics are blocked by your
consent settings
Deep Generative Model-based RSSI Synthesis for Indoor Localization
(2022-01-01)
Suroso, Dwi Joko
;
Cherntanomwong, Panarat
;
Sooraksa, Pitikhate
Indoor localization via deep learning (DL) is attracting researchers' attention. DL is mainly used for fingerprinting-based indoor localization as it generally employs a vast offline database to ensure its reliability. However, the long effort and high cost of constructing this database are the disadvantages of this technique. This paper implements variational autoencoders (VAE), one of the popular deep generative models, to alleviate the drawbacks of offline database issues. Our proposal works using the received signal strength indicator (RSSI); unfortunately, it is known for its fluctuation and instability. Thus, instead of using RSSI directly as a localization parameter, we learn its distribution via VAE to generate the synthetic RSSI values. We utilized the RSSI from an actual measurement campaign. The VAE implementation results show that we can obtain the RSSI synthesis by exploring the latent distribution learned from the input distribution. Thus, the offline database density grids can be enhanced. We validated the results by varying epochs to map the learned latent distribution. However, we still have relatively low accuracy in the synthetic RSSI values, especially when applying a small number of epochs, i.e., 10 and 100. When we applied epoch number 1000, the error was relatively low (-3dBm average error) in the sampled position. Our preliminary assumption is that the dataset is small for VAE learning, and probably the 3-by-3 RSSI-to-image size assumption could still be inadequate.
Some of the metrics are blocked by your
consent settings
A deep neural network-correlation phase sensitive mask based estimation to improve speech intelligibility
(2023-09-01)
Sivapatham, Shoba
;
Kar, Asutosh
;
Bodile, Roshan
;
Mladenovic, Vladimir
;
Sooraksa, Pitikhate
General masking-based speech enhancement using a deep learning architecture (DNN) approach focuses on the spectral values of the speech in order to show improvement in intelligibility. But, the residual noise present in the phase spectrum and inter-channel correlation dependency between noise and noisy speech can impact the results of intelligibility in speech enhancement. This research work proposes a correlation phase-sensitive novel mask which contains phase, magnitude spectral and inter-channel correlation for a tangible improvement in speech intelligibility. The correlation parameter finds the dependency between the signals and phase spectrum factor eliminates the residual noise. In addition, selecting the prior features from the feature combination also helps in reducing the dimensionality and increases the accuracy of the enhanced speech. This work also aims to decrease the complexity of the DNN by analysing the network with different parameters. The performance of the mask is evaluated with various intelligibility factors. The proposed mask has been compared with different mask estimators. The proposed mask has generated estimated speech with an increase in the intelligibility of 0.02-0.001 over six different noises and four different signal-to-noise (SNR) levels.
Some of the metrics are blocked by your
consent settings
3D range-based indoor localization by using only two beacons
(2023-05-22)
Suroso, Dwi Joko
;
Krisnawan, Aditya Bagus
;
Sooraksa, Pitikhate
;
Cherntanomwong, Panarat
Indoor localization has been active research in the past decade. The lack of a general model and the demand for high accuracy push the research forward. Some researchers have proposed unique variations and combinations to answer the challenges from the technologies to techniques. Some publications considered the low-cost equipment to the simple algorithm or model to the advanced, robust, and sophisticated system. Almost all of them use the two basics technique; range-based and range-free techniques. Each technique has advantages and disadvantages, including how the technique can efficiently handle the indoor multipath propagation effects. Most of the papers published are also considered the same room or single room indoor environment as the indoor localization system measurement campaign. This paper proposes a three-dimensional (3D) indoor localization for multi-story buildings using simple technology and technique and using only two beacons. We utilized the received signal strength indicator (RSSI) from ESP8266 based on the Wireless-Fidelity (Wi-Fi) standard as the localization parameter. We employ the range-based technique of min-max and least-square. We also compare both techniques to observe which one is suitable for the multi-story building implementation. We also emphasized the challenge of using only two beacons as our contribution. Our system performance results show that the mean error accuracy for min-max is 1.93m, while the least-square yields the mean error of 5.48m. The minimum error of min-max and least-square are 0.33m and 0.77m, respectively. The maximum error of least-square can reach more than 10m, while the min-max gives the maximum error of 4.38m. The results prove that RSSI-based min-max can be applied in a particular condition using only two beacons in the multi-story building.
Some of the metrics are blocked by your
consent settings
A comprehensive review of flood-prone area zonation using ensemble and hybrid machine learning models with a framework proposal for modelling
(2025-01-01)
Long, Gen
;
Tantanee, Sarintip
;
Nusit, Korakod
;
Sooraksa, Pitikhate
This research offers a systematic review encompassing 63 relevant peer-reviewed papers concerning flood susceptibility, hazard and risk assessment using various ensemble machine learning and hybrid approaches. It examines publication details, study characteristics, terminology, flood inventories, conditioning factors, data resolution, and modeling approaches. A key contribution is a proposed framework for practising ensemble or hybrid modeling. The framework comprises data preparation, checking for multicollinearity, factor selection and weighting, optional factor optimization, k-fold cross validation where appropriate, ensemble or hybrid modeling, and model evaluation. This framework aims to facilitate research activities and enhance model quality. Furthermore, the statistical outcomes can benefit researchers by guiding their further research. The knowledge produced by this study will thus help advance understanding of the application of ensemble machine learning and hybrid methods for the zonation of flood-prone areas and guide the direction of further research on enhancing the effectiveness of flood risk management strategies.
Some of the metrics are blocked by your
consent settings
Synthesis of a Small Fingerprint Database through a Deep Generative Model for Indoor Localisation
(2023-01-01)
Suroso, Dwi Joko
;
Cherntanomwong, Panarat
;
Sooraksa, Pitikhate
In deep learning (DL), the deep generative model is helpful for data augmentation objectives to tackle the lack of datasets that have a significant impact on learning performance. Data augmentation or synthesis is expected to solve the issue in a small/sparse database. The problem of databasing also exists in the fingerprint-based indoor localisation system. The dense offline fingerprint database must be constructed with the accuracy requirement. However, this will affect the high cost, massive laborious work, and increase the complexity of the system. Therefore, this paper proposes to address these issues by generating synthetic data via a deep generative model. The generative adversarial network (GAN) is selected to generate the synthetic fingerprint database for indoor localisation. Our database consideration consists of power-based parameters, i.e., the received signal strength indicator (RSSI) from Wi-Fi devices obtained from the actual measurement campaign. Some of the literature mainly discusses how GAN works in a vast and complex dataset. Here, we consider applying GAN in a relatively small dataset and for a simple setup. Our results show that by only using the 20 % fraction of actual RSSI data combined with the synthetic RSSI, the accuracy validation performance is slightly higher than when using all actual data usage. Moreover, in only 60 % of actual data usage and in combination with 625 samples of synthetic data, the accuracy performance is improved to 0.73 (1.37 times higher than the use of all actual data, 0.53). Thus, this result proves that the challenges of offline fingerprint databases can be alleviated by data synthesis through GAN by using only a small dataset.
Some of the metrics are blocked by your
consent settings
Development of salt-sensing system for jellyfish desalting process
(2020-01-01)
Rakmae, Samak
;
Sooraksa, Pitikhate
;
Pornchaloempong, Pimpen
In the seafood industry, a large amount of salt is added to preserve seafood products. During processing, it is necessary to desalt the products for the sake of customers' health. Unlike a large-scale factory, many small enterprises lack tools and methods for desalting and measuring the efficiency of the desalting process. We have developed a rapid prototype salt-sensing system that can measure the desalting efficiency. Dried salted jellyfish are used as testing materials to evaluate the system. The rapid prototype comprises a microcontroller, a wire, and a liquid crystal display. Using a simple mapping between electrical conductivity and actual data obtained from the measurement of samples, the sensing system is successfully calibrated. A method of desalting the salted jellyfish material is also proposed. This desalting method and the newly developed simple sensing system for the desalting process are expected to make a significant contribution to the seafood processing industry.
Some of the metrics are blocked by your
consent settings
Surface Temperature Limit as Food Quality Control in Automatic Learning Model for Drying Process
(2021-04-01)
Pongsuttiyakorn, Thadchapong
;
Sooraksa, Pitikhate
;
Pomchaloempong, Pimpen
This paper presents design and implementation of an automatic learning model for a drying process. Setting surface temperature limit as an upper boundary for the drying process is very helpful key to prevent loss in physico-chemical properties such as color variation, nutrients, preferable odors, and surface textures. Based upon input-out data acquired from the designed system, the drying machine can identify system parameter adjusting by innovation sequences of the Kalman gains. The model is then used as a predictor to prescribe suggestion rules for firing automatically suitable control gains. According to the experimental results, Thai curry paste as testing materials under the proposed control process reveal desired properties, meaning that the scheme is effective and is available to be adopted for other similar dried food requirements.
Some of the metrics are blocked by your
consent settings
Flood Susceptibility Mapping Using Machine Learning Models with Novel Flood Inventory Sampling Strategies
(2025-01-01)
Nusit, Korakod
;
Tantanee, Sarintip
;
Sooraksa, Pitikhate
In this study, we introduce an innovative frequency-area-weighted sampling method to address spatial and temporal biases in flood inventory creation. Focusing on Thailand’s Nan River Basin, we integrated 13 flood conditioning factors and developed a point-based inventory that includes 3000 flood and 3000 non-flood samples, proportionally allocated on the basis of flood recurrence intervals and spatial distribution. We evaluated four machine learning models—artificial neural network, support vector machine, K-nearest neighbors, and random forest (RF) models—to assess their performance in flood susceptibility mapping (FSM). Among these, the RF model demonstrated the highest predictive capability, achieving an area under the curve (AUC) of 0.979 for the test set and an AUC of 0.984 for the verification set. The resulting susceptibility map identified 10.64% of the study area as “very high” risk, providing critical insights for prioritizing flood mitigation efforts. This work advances FSM methodology by effectively bridging the temporal flood frequency and spatial heterogeneity in inventory design, offering a robust framework for data-driven flood risk management in vulnerable regions.
Some of the metrics are blocked by your
consent settings
Eye-hand Coordination Simulator of Robot Arms for Science, Technology, Engineering, and Mathematics Education
(2025-01-01)
Preedanont, Pimpran
;
Sooraksa, Pitikhate
In this paper, we present an eye-hand coordination simulator for robot arms as a compact cyber-physical science, technology, engineering, and mathematics (STEM) learning unit that links visual perception to robot motion in pick/place interactions with a mobile robot as an automatic guided vehicle (AGV). The unit integrates four domains into one workflow: science (kinematics and motion), technology (sensors, motor controllers, vision), engineering (mechanisms and control states), and mathematics (geometric computation and frame transforms). The pick/place machine prototype includes linear X-Y-Z-axes with rotary and flip joints to realign an item box between a shelf and an AGV. A vision system detects a pair of fiducial circles to estimate the AGV centerline, yaw, and slot positions, while displacement sensors measure stand-off and assist parallel alignment. Performance was evaluated using a mock-up AGV positioned with varied offsets and yaw within a ±10 mm parking tolerance. Across 10 trials, the vision-based estimates of middle-slot X and stand-off Sx closely matched tape measurements, achieving 98-100% accuracy. The results show that the simulator is dependable in vision-guided coordination and usable as a simple, accessible platform for STEM education.

Sooraksa, Pitikhate

Filters

Settings

Sort By

Results per page