KPKhem Narayan Poudyal

Khem Narayan Poudyal

*This is an algorithmically generated author record based on publicly available information.

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article

Previous Organizations / Institutes

The organizations / institutes from the scholar's other articles in the past ten years


Khem Narayan Poudyal has a total of 40 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Open Access

    Reinforcement Learning-Based Autonomous Soccer Agents: A Study in Multi-Agent Coordination and Strategy Development

    by Biplov Paneru, Bishwash Paneru, Ramhari Poudyal, Khem Poudyal

    Reinforcement learning (RL) approaches, particularly Q-learning, have emerged as strong tools for autonomous agent training, allowing agents to acquire optimum decision-making rules through interaction with their surroundings. This research investigates the use of Q-learning in the context of training autonomous agents for robotic soccer, a complex and dynamic arena that necessitates strategic planning, coordination, and adaptation. We studied the learning progress and performance of agents taught using Q-learning in a series of experiments carried out in a simulated soccer setting. During training, agents interacted with the soccer environment, iteratively changing their Q-values in response to observable rewards and behaviors. Despite the high-dimensional and stochastic character of the soccer domain, Q-learning helped the agents develop excellent tactics and decision-making capabilities. Notably, our study found that, on average, the agents required 64 steps to reach a stable policy with an average reward of -1.


  • Open Access

    Comparative Analysis of Different Models within RadEst 3.0 for Solar Radiation Estimation at Dhankutta, Nepal

    by Basanta Kumar Rajbanshi, Ram Gopal Singh, Bed Raj Kc, Khem Narayan Poudel

    To develop efficient solar energy technologies, it is essential to understand accurately the distribution of solar radi- ation in each geographical location. This research project estimates total solar radiation reaching a surface daily in Dhankutta (26.983°N, 87.346°E, 1192 m), Nepal by utilizing weather data including temperature, humidity, wind speed, rainfall, and measured solar radiation data. Four distinct models (Bristow-Campbell (BC), Campbell-Donatelli (CD), Donatelli-Bellocchi (DB), and Donatelli-Campbell-Bristow-Bellocchi (DCBB)) were employed to estimate global solar radiation (GSR). Root mean square error (RMSE), coefficient of residual mass (CRM), mean bias error (MBE), mean percentage error (MPE), and coefficient of determination (R2 ) were among the statistical tools used to assess the effectiveness of models. By minimizing RMSE, and CRM, and maximizing R2 , parameter fitting (PF) is used to calibrate all four models. The annual mean daily GSR for 2021 and 2022 was determined to be 14.5 ± 0.25 MJ/m2/day and 15.7 ± 0.24 MJ/m2/day, respectively, indicating sufficient potential for solar energy generation. The maximum GSR of 37.6 MJ/m2 /day and 26.5 MJ/m2 /day were observed in 2021 and 2022 respectively. Among the four models used to estimate GSR, the CD model showed the highest accuracy with an R2 value of 0.60 so the CD model is considered the most reliable method for predicting GSR levels in this area.


  • Open Access

    Autism Spectrum Disorder Prediction Using Hybrid Deep Learning Model and a Recommendation System Application for Autistic Patient

    by Biplov Paneru, Bishwash Paneru, Krishna Bikram Shah, Awan Shrestha, Ramhari Poudyal, Khem Narayan Poudyal

    The goal of this study is to create machine learning models that use a big dataset to predict Autism Spectrum Disorder (ASD). To achieve optimal performance, a number of algorithms were employed and refined, including Support Vector Machines (SVM), Random Forest, XGBoost, Multi-Layer Perceptron (MLP) and a hybrid model that combines MLP and SVM. To evaluate the durability of the model's performance, the study used cross-validation and hyperparameter tuning techniques. Measures such as memory, accuracy, precision and F1-score have been employed to assess how well the models predict ASD. It's interesting to note that the RBF kernel did quite well in the grid search using the SVM model. All models produced good findings, with test set accuracies ranging from 87-97%. With 97% accuracy on the testing set, the CatBoost algorithm demonstrated excellent performance. Additionally, the hybrid MLP + SVM model demonstrated the potential benefits of combining different approaches by doing well on both the training and testing sets. Additionally, a Flask application was made to provide a straightforward user interface for the machine learning models that were learned. For those with ASD or who are at risk, this application generates predictions based on user input and provides tailored recommendations and interventions. The work highlights the potential for developing useful tools to support ASD diagnosis and intervention, as well as the effectiveness of machine learning techniques in ASD prediction. The robustness and applicability of the existing models may be strengthened by more research and validation on bigger and more varied datasets


  • Open Access

    Satellite-pose estimation using IMU sensor data and Kalman filter with RF-433 Mhz powered communication and helical antenna design for ground station

    by Biplov Paneru, Ankit Adhikari, Bishwash Paneru, Krishna Bikram Shah, Sanjog Chhetri Sapkota, Ramhari Poudyal, Khem Narayan Poudyal


  • Open Access

    Wavelet Analysis of Atmospheric Ozone and Ultraviolet Radiation on Solar Cycle-24 over Lumbini, Nepal

    by Prakash M. Shrestha, Suresh P. Gupta, Usha Joshi, Morgan Schmutzler, Rudra Aryal, Babu Ram Tiwari, Binod Adhikari, Narayan P. Chapagain, Indra B. Karki, Khem N. Poudyal

    This research aims to comprehensively examine the clearness index (KT), total ozone column (TOC), and ultraviolet A (UVA) and ultraviolet B radiation (UVB) over Lumbini, Nepal (27°28’ N, 83°16’ E, and 150 m above sea level) throughout the 11 years of solar cycle 24 (2008 to 2018). The Lumbini, a highly polluted region, is important in advancing the identification and analysis of TOC variations across regions with similar geographical and climatic attributes. Data from the Ozone Monitoring Instrument (OMI) of the EOS-AURA satellite of NASA were used to analyze the daily, monthly, seasonal, and annual trends in the clearness index (KT), ultraviolet A (UVA), ultraviolet B (UVB), and TOC from the Comprehensive Environmental Data Archive (CEDA). The study found that the yearly averages for KT, TOC, UVA, and UVB were 0.55 ± 0.13, 272 ± 14 DU, 12.61 ± 3.50 W/m2, and 0.32 ± 0.11 W/m2, respectively. These values provide insights into the long-term variations in atmospheric parameters at Lumbini. The study also applied the continuous wavelet transform (CWT) to analyze KT, TOC, UVA, and UVB temporal variations. The power density peak of 35,000 DU2 in the TOC was observed from the end of 2010 to the end of 2011, within 8.5 years, underscoring the significance of analyzing TOC dynamics over extended durations to understand atmospheric behavior comprehensively.


  • Open Access

    Automated Environmental Stewardship: A Ribbon-Cutting Robot with Machine Vision for Sustainable Operation

    by Biplov Paneru, Bishwash Paneru, Ramhari Poudyal, Krishna Bikram Shah, Khem Narayan Poudyal, Yam Krishna Poudel

    This paper provides a novel way for automating ribbon-cutting rituals that use a specifically constructed robot with superior computer vision capabilities. The system achieves an outstanding 92% accuracy rate when assessing picture data by using a servo motor for ribbon identification, a motor driver for robot movement control, and nichrome wire for precision cutting. The robot's ability to recognize and interact with the ribbon is greatly improved when it uses a Keras and TensorFlow-based red ribbon identification model which obtained accuracy of about 93% on testing set before deployment in system. Implemented within a Raspberry Pi robot, the method exhibits amazing success in automating ceremonial activities, removing the need for human intervention. This multidisciplinary method assures the precision and speed of ribbon-cutting events, representing a significant step forward in the merging of tradition and technology via the seamless integration of robots and computer vision.

  • Open Access

    Green Energy Production Aid Spider Robot: An Innovative Approach for Waste Separation Using Robotic Technology Powered with IoT

    by Biplov Paneru, Bishwash Paneru, Krishna Bikram Shah, Ramhari Poudyal, Khem Narayan Poudyal

    Growing worldwide waste management challenges have prompted research efforts for sustainable solutions, especially in green energy technologies. This study highlights robotics’ critical role in modern waste management by putting forth a novel strategy incorporating cutting‐edge technologies. The research aims to create an effective waste monitoring system to measure the amounts of methane gas and organic waste at disposal sites. The main component of this method is a 3D‐printed spider robot driven by a servo motor with solid and flexible claws. This robotic system communicates gas level readings to the ThingSpeak platform on the Internet of Things, while it navigates waste disposal sites autonomously. The study also investigates how to develop robotic systems more economically and efficiently by using a portable, low‐power microcontroller, such as the Arduino Nano. The study uses a rigorous testing methodology to assess the performance and viability of the microcontroller and robotic system. The results are expected to emphasise a practical and economic development approach while offering insightful information on developing sustainable waste management technologies.


  • Open Access

    An Assessment of the Impact of Land Use and Land Cover Change on the Degradation of Ecosystem Service Values in Kathmandu Valley Using Remote Sensing and GIS

    by Srijana Shrestha, Khem Narayan Poudyal, Nawraj Bhattarai, Mohan B. Dangi, John J. Boland

    Land use and land cover (LULC) robustly influence the delivery of the ecosystem services that humans rely on. This study used Kathmandu Valley as a study area which is a fast-growing and most vulnerable city to climate change. Remote sensing and GIS methods are the most significant methods for measuring the impact of LULC on the ecosystem service value (ESV). The satellite-based dataset was used for quantitative assessment of the LULC and ecosystem service value for 10-year intervals from the year 1989 to 2019. The result revealed that the area of forest cover, cropland, and waterbodies decreased by 28.33%, 4.35%, and 91.5%, respectively, whereas human settlement and shrubland increased by more than a hundred times and barren land by 21.14% at the end of the study period. This study found that Kathmandu valley lost 20.60% ESV over 30 years which dropped from USD 122.84 million to USD 97.54 million. The urban growth and extension of agricultural land to forest cover areas were found to be contributing factors for the reduction in ESV of Kathmandu valley. Cropland transformed into shrubland, bringing about an increase in ESV of some areas of the study region. In conclusion, the aggressive increase in population growth with inadequate urban planning and fragmentation of farmlands influenced the ESV of Kathmandu valley.


  • Open Access

    Estimation of Solar Insolation and Angstrom–Prescott Coefficients Using Sunshine Hours over Nepal

    by U. Joshi, P.M. Shrestha, S. Maharjan, A. Bhattarai, N.P. Chapagain, I.B. Karki, K.N. Poudyal

    The amount of solar insolation that reaches the Earth in one hour is sufficient to fulfill its annual energy budget. One of the challenges for harvesting this energy is due to a lack of relevant data. In the least developed countries like Nepal, the number of observation stations is insufficient. This data gap can be filled by employing credible empirical models to estimate solar insolation in regions where insolation measurements are not available. In this paper, AngstromPrescott model parameters are estimated for fifteen different locations of Nepal. Then, correlation is developed for the prediction of solar insolation using only sunshine hour data. The different statistical parameters such as root mean square error (RMSE=1.958), mean bias error (MBE=0.018), mean percentage error (MPE=2.973), coefficient of residual mass (CRM=0.001), and correlation coefficient (r=0.909) were used to validate the developed coefficients. The resulting AngstromPrescott coefficients are a=0.239 and b=0.508. These coefficients can be utilized for the prediction of solar energy at different parts of the country in similar weather conditions.