Browse Topic: Machine learning

Items (1,496)
Find
EKF-Based Machine Learning Approach for Battery SOH Prediction in Commercial Vehicles Fleet2026-24-0002To be published on 09/21/2026
Estimating battery State of Health (SOH) from field data is essential to ensure successful operation and to increase the uptime of battery electric vehicles (BEVs). Most studies in the literature have proposed several methods for battery SOH estimation, relying on datasets acquired under controlled laboratory conditions. However, SOH estimation becomes significantly more challenging when dealing with real-world driving data due to the increased variability and complexity of operating conditions. In this work, CAN telematics data, sampled at 1 Hz, have been collected across approximately 16 months of operation from 10 electric commercial vehicles. During this period, a battery degradation of up to 2.5 - 3% is observed within the fleet. Firstly, a model-based framework is introduced, in which a battery second-order Equivalent Circuit Model (ECM) has been coupled with an Extended Kalman Filter (EKF) to estimate the battery SOH. The results confirmed that the EKF accurately captures the
D'Agostino, Valerio, Pulvirenti, Luca, Shanker, Anirudh, Cardone, Massimo, Rizzoni, Giorgio, Vitale, Francesco
Multi-sensor Data Fusion Based on Big Data Modelling2026-99-0568To be published on 07/10/2026
The collection of road high-frequency data often involves inputs from multiple sensors, such as stress and strain, and sampling of these data features a high sampling rate of up to 2,000 Hz. High-frequency sampling enables capturing of the internal stress and strain of the pavements when vehicles are passing and facilitates the analysis of the pavement structure and prediction of its long-term service performance. However, while the sensors are continuously collecting data, the time the vehicles pass is discrete and unpredictable, resulting in a large number of low information density or irrelevant data. Even when the massive high-frequency data are collected, challenges remain in data transmission, storage, and analysis—the challenges are attributable not only to the massive quantity and complexity of data from multiple sensors, but also to the inconsistent data formats, misaligned timestamps, and multi-sensor data fusion difficulties. In response to the challenges specified above, a
Gang, Jian, Zhang, Yue, Chen, Yinghao, Zheng, Xiaoyan, Wang, Taojie, Liu, Yilin, Guan, Wei, Wu, Jiangfeng
Traffic Flow Prediction System Based on Deep Learning2026-99-0549To be published on 07/10/2026
Traffic flow prediction is of great significance for improving the operation efficiency of the transportation system, optimizing travel experience and reducing traffic congestion. Traditional traffic flow prediction methods are difficult to capture the spatio-temporal nonlinear characteristics of traffic flow due to its simple model and insufficient feature extraction ability. Therefore, an intelligent traffic flow prediction system based on deep learning is proposed, constructs a deep learning model based on graph convolution and fusion of attention mechanism LSTM. Based on this, a traffic flow prediction system is implemented. Experiments show that, on the PeMSD4 and PeMSD4 datasets, the error of the model in RMSE and Mae indicators is significantly reduced compared with the traditional methods, which provides an efficient solution for traffic flow prediction and congestion analysis, and has both theoretical innovation and engineering practical value.
Tang, Zhan, Lu, Xiaoyu, Yang, Nian, Xiang, Xiaohong, Hou, Xiang, Peng, Xiaoli
A Review of UAV Support Technologies for Low-Altitude Airspace Operation and Management2026-99-0597To be published on 07/10/2026
In recent years, with the low-altitude economy developing rapidly, the operation and management of low-altitude airspace has gradually become a hot topic. Unmanned aerial vehicles (UAVs) constitute a fundamental component of the low-altitude airspace ecosystem, significantly influencing its structure and functionality. The technological advancement of UAVs has fundamentally transformed the operational paradigm for low-altitude airspace management. This paper presents a comprehensive review of UAV-supported technologies in the context of low-altitude airspace operations and management. It systematically analyzes key technologies and applications of UAVs in areas such as airspace capacity and safety assessment, trajectory planning, and standardized flight management. Drawing from kinematic analysis and traffic flow theory, UAV density control and collision risk prediction offer quantitative insights into airspace capacity evaluation. Additionally, probabilistic analysis and simulation
Gong, Lei, Ma, Zhenxiao, Luo, Qin
Research on Urban Bus Timetable Optimization Based on Deep Reinforcement Learning2026-99-0538To be published on 07/10/2026
Public transportation serves as a crucial component of urban mobility, contributing to the alleviation of urban congestion, reduction of travel expenses, and mitigation of air pollution. Nonetheless, the dynamic passenger demand and the complex traffic conditions render traditional bus timetables inadequate, leading to ineffective allocation of public transportation resources. Consequently, it is essential to create bus timetables that are responsive to actual traffic scenarios and fluctuating passenger demand. This study regards the bus timetable planning problem as a Markov decision-making process within a discrete time framework, proposing a deep reinforcement learning-based optimization model for bus timetables. In particular, the model is designed to account for both bus companies and passengers, incorporating a state space and reward calculation method that emphasizes passenger comfort. Then Deep Q-Network (DQN) methodology is employed to issue instructions on whether a bus
Xu, Jie, Xia, Dong, Yang, Jianxi, Wang, Bing
Traffic Crash Prediction and Causal Disentanglement by Machine Learning2026-99-0552To be published on 07/10/2026
Aiming at the problem of insufficient modeling of spatio-temporal heterogeneity in road traffic accident prediction, a dual task machine learning framework integrating geographical environment, location attributes and time periodicity is proposed. The dataset used in this study was derived from traffic accident records of Nanchang during 2019–2023. Firstly, geographical identifiers are generated by rounding and aggregating latitude and longitude coordinates. At the same time, the location type is processed by a one-hot encoding, so as to carry out spatial clustering analysis of accident hotspots. Compared with the North-South pattern, the contribution of geographical features shows a strong East-West trend. The kernel density heatmap identified Zone A and zone B as dual core high-risk areas. Secondly, the sinusoidal/cosine function is used to encode the time feature circularly, which effectively captures the daily change of the accident. The quantitative analysis of random forest
Luo, Jiang, Zhang, Yuxin, Li, Xin, Wu, Ronghai
Research on Predicting Occupant Height Using CNN-Based Pressure and Posture in Zero-Gravity Seats2026-99-0521To be published on 07/10/2026
Zero-gravity seats alleviate prolonged sitting fatigue by optimizing human body pressure distribution, but the correlation mechanism between body size parameters and pressure distribution remains unclear. This study proposes a deep learning model based on multimodal data fusion, combining pressure matrices and postural angle data to construct a convolutional neural network (CNN) with a height prediction error ⩽3 cm. Experiments collected pressure and posture data from 100 participants with diverse anthropometric percentiles. Through the fusion of features and the optimization of the model, the study managed to quantify how height and weight impact pressure gradients. The results indicate that the model achieved a prediction R2 value of 0.73, which confirms that there is a strong correlation between pressure distribution and body size parameters. The findings offer theoretical and technical support for the adaptive adjustment systems within intelligent cabins.
Bi, Tengfei, Nie, Jiacheng, Du, Changjiang, Ji, Yuechen, Wang, Song, Sun, Jiawei
An Application-Oriented Process Model for Selecting Uncertainty Quantification Methods in Machine Learning2026-01-07737/1/2026
Uncertainty quantification (UQ) is increasingly recognized as essential when machine learning (ML) is employed in domains that are safety-relevant, cost-intensive, or legally binding, such as the product engineering of battery electric vehicle (BEV) energy systems. UQ methods aim to estimate the aleatoric, epistemic or both uncertainties associated with the predictions of a machine learning model. However, the landscape of UQ methods is diverse and rapidly evolving, with no single approach proving optimal across all tasks. Consequently, the selection of methods in practice is often driven by experience, constrained by limited comprehensive knowledge, time, and implementation capacity. This paper introduces an application-oriented process model supporting data scientists in selecting UQ methods in ML by adapting the SPALTEN [1] problem-solving methodology and the Algorithm Selection Process Model (ASPM) into an Algorithm Selection Process Model for Uncertainty Quantification (UQ-ASPM
Holderied, Niklas, Hörtling, Stefan, Bause, Katharina, Düser, Tobias
Development of a Digital Twin of an Electrical Powertrain in a Testbench Environment for Efficiency Estimations2026-01-07447/1/2026
The aim of this work is to develop a modular, real-time-capable digital twin of an electric powertrain based on machine learning (ML)-based model structures and a systematic, component-oriented architecture with a focus on efficiency estimation in test bench environments. The further goal here is to enable virtual testing, which can be used for frontloading and thus both prevent errors and increase the speed of product development. Based on a comprehensive set of measured and derived test bench data, a multi-stage procedure is implemented that integrates data acquisition, physically informed feature selection, modeling at the component and subsystem level, and hybrid coupling strategies. The digital twin captures inverter, electric machine, and mechanical transmission stages and generates consistent predictions of key variables such as torque, speed, power factors, and subsystem as well as overall drivetrain efficiency. The methodology enables a systematic comparison of black box, dark
Kopp, Lennart, Proksch, Daniel, Ockert, Niels, Karthaus, Carsten, Kley, Markus
Production-Ready Automated ECU Calibration Using Residual Reinforcement Learning2026-01-07807/1/2026
Electronic Control Units (ECUs) have played a pivotal role in transforming motorcars of yore into the modern vehicles we see on our roads today. They actively regulate the actuation of individual components and thus determine the characteristics of the whole system. In this, the behavior of the control functions heavily depends on their calibration parameters which engineers traditionally design by hand. This is taking place in an environment of rising customer expectations and steadily shorter product development cycles. At the same time, legislative requirements are increasing while emission standards are getting stricter. Considering the number of vehicle variants on top of all that, the conventional method is losing its practical and financial viability. Prior work has already demonstrated that optimal control functions can be automatically developed with reinforcement learning (RL); since the resulting functions are represented by artificial neural networks, they lack
Kampmeier, Andreas, Badalian, Kevin, Koch, Lucas, Lee, Sung-Yong, Andert, Jakob
Interpretable Tire Force Modeling for Formula Student Vehicle Dynamics and Lap Time Applications2026-01-07627/1/2026
Accurate tire models are a key enabler for vehicle dynamics simulation, control design, and lap time optimization, particularly in the context of Formula Student race cars, where vehicle setups and tire characteristics differ significantly from production vehicles. State-of-the-art tire models, such as Pacejka’s Magic Formula, generally provide high prediction accuracy. However, their predefined functional structure and large number of coupled parameters are designed for broad applicability across many tire types rather than for specific racing tires. This often results in limited interpretability, nontrivial parameter identification, and unnecessary model complexity for specialized applications such as Formula Student. This paper presents a data-driven approach for deriving compact and physically interpretable tire force models using symbolic regression. The proposed method employs an intelligent tree search to systematically explore the space of mathematical expressions and identify
Anselment, Marcel, Borowski, Julian, Rudolph, Stephan
Who’s Riding? Machine Learning-Powered Rider Identification from E-Bike CAN Data2026-01-07787/1/2026
This study investigates the feasibility of identifying individual e-bike riders based on CAN data using machine learning techniques. Datasets from 12 test riders performing various predefined cycling tasks on a dynamometer test bench are collected and used to ensure controlled and reproducible conditions. The recorded CAN data includes various sensor signals, such as power output, cadence, torque, and the used support mode. After pre-processing, two different methods of feature extraction are tested and compared, one based on snapshots of the data and one based on driving events such as braking and accelerating, measured by calculating statistics of the riding data over sliding windows. A range of machine learning models is employed to classify riders based on their distinct riding patterns using the extracted features. The evaluated models comprise KNN, Random Forest and Naïve Bayes. The findings demonstrate the efficacy of machine learning in differentiating riders, with Random
Simmann, Gabriel, Rauch, Yannick, Beißert, Florian, Kriesten, Reiner
BO-NRPPO: A Bayesian Optimization–Tuned Proximal Policy Optimization Framework for Semi-Active Suspension Control10-10-03-00256/25/2026
Semi-active suspension systems enhance ride comfort and handling performance by adaptively modulating damping characteristics. However, conventional model-based controllers often fail to maintain optimal performance under uncertain and time-varying vehicle conditions. This article proposes Bayesian Optimization–Tuned Proximal Policy Optimization with Non-Parametric Rewards (BO-NRPPO), a novel reinforcement learning (RL) framework that integrates Bayesian Optimization (BO) with Proximal Policy Optimization (PPO) and a non-parametric reward function (NRF). The proposed approach enables adaptive self-tuning, data-driven reward shaping, and uncertainty-aware policy learning. Moreover, a Trapezoidal Simple Moving Average (TSMA)–based reward normalization scheme is introduced to accelerate convergence and stabilize training. Simulation results across diverse driving scenarios demonstrate that BO-NRPPO outperforms the passive suspension, the classical Linear Quadratic Regulator (LQR), and PPO
Chen, Guoying, Wang, Xinyu, Wang, Jiaqi, Zhan, Xinwang, Bi, Chenxiao, Cong, Shiqi, Hua, Min, Sun, Tianjun, Gao, Zhenhai
Best-in-Class NVH Optimisation for Electric Powertrains, Holistic Front-Loading Approaches2026-01-06896/20/2026
Achieving best-in-class Noise, Vibration, and Harshness (NVH) in electric powertrains demands a paradigm shift in development methodology. This paper presents a practice-oriented overview of simulation methods in NVH development methodology for electric drive units. This includes target cascading and multi-objective optimisation, and by attacking NVH at the source using KPIs early in the design cycle, significant reductions in development time and reliance on traditional testbed loops are realised. Machine learning (Neural Network) algorithms are utilized to find the best-in-class design, using multi-objective optimisation as well as refining simulation accuracy by adding tolerance effects while target cascading ensures alignment of system-level performance objectives down to subsystem contributions. Combined, these strategies enable rapid and robust NVH optimisation, using simulation for next-generation electric powertrain development. Several applications and real-life examples
Mehrgou, Mehdi, Garcia de Madinabeitia, Inigo, Graf, Bernhard, Gojo, Josef
NoiseSphere: An AI-Based Approach for Dynamic and Large-Scale Noise Mapping2026-01-07196/20/2026
Noise pollution is a major environmental and health challenge, yet its strong spatial and temporal variability makes comprehensive mapping highly complex. Current approaches under the European Noise Directive (END) provide only partial coverage and often lack temporal dynamics. The NoiseSphere project, funded by the Austrian Research Promotion Agency FFG, develops an AI-based methodology for dynamic, large-scale noise prediction and mapping. A machine learning model is trained on heterogeneous data sources, including semantically enriched open Sentinel-2 satellite imagery, OpenStreetMap road data and existing noise maps. The model is refined through integration of noise emission data and validated using targeted in-situ measurements. A case study in an urban environment (Graz, Austria) demonstrates the model’s applicability. By combining remote sensing, traffic dynamics, and machine learning, NoiseSphere enables predictive noise mapping even in regions not covered by current
Girstmair, Josef
Active Learning Supported Metadata Extraction from NVH Simulations2026-01-07206/20/2026
Simulations can only be searched, reused and leveraged as training data for machine learning methods if suitable metadata are related. Manually obtaining these metadata is time-consuming and requires expert knowledge. Consequently, there often is a lack of metadata and this prohibits the reutilization of simulation data. Therefore, automated frameworks for metadata extraction are essential to obtain metadata information quickly, effortlessly and cost-efficiently. At present, there are no toolboxes for Finite-Element-Simulation data. Nevertheless, machine learning methods are a promising solution for this task. Training classical supervised machine learning methods for metadata generation often faces the lack of labeled data since manual labelling can be very costly. Therefore, rule-based extraction algorithms are used as an alternative for fundamental metadata extraction. For more enhanced tasks they are often not feasible. Active Learning is a suitable technique to overcome this
Luegmair, Marinus, Gröttrup, Sören
A Self-Learning Framework for NVH CAE Analysis2026-01-07186/20/2026
The vibro-acoustic performance of a vehicle is a critical factor in customer perception of quality and comfort, yet optimizing for Noise, Vibration, and Harshness (NVH)—specifically road noise—presents a persistent challenge in the modern automotive development cycle. While advanced Finite Element Method (FEM) analysis is essential, the increasing complexity and volume of CAE simulation data often overwhelm manual interpretation, potentially leading to prolonged development times or compromises in final comfort quality. To address these challenges, this paper introduces the application of CDH/ACE (Autonomous Computational Experiments), a framework that integrates conventional CAE simulation workflows with advanced machine learning in an iterative, cyclic process. This creates an exceptionally user-friendly and self-correcting system that autonomously defines, performs, and learns from computational experiments. By leveraging machine learning algorithms to build robust predictive models
Visser, Rene
Unsupervised Feature Frequency Map Extraction for Simple C-EPS Defect Detection2026-01-07176/20/2026
In recent years, the automotive industry has actively explored the application of various AI-based models such as Convolutional Neural Networks (CNN), Long Short-Term Memory (LSTM) networks, Autoencoders, and Transformers to improve defect detection rates at the End-of-Line (EOL) stage. However, implementing these approaches in the Noise, Vibration, and Harshness (NVH) area face several practical challenges: ① extended evaluation times compared to other data types, which limit the quantity of training data and lead to overfitting; ② label imbalance caused by the relatively small amount of defect data; ③ reduced labeling accuracy due to human error; ④ decreased robustness under domain shifts such as changes in jig fixtures, test environments, and signal-to-noise ratio (SNR); ⑤ diminished model reliability when new defect arise during development; and ⑥ constraints imposed by compatibility requirements with existing test equipment. This study proposes a Convolutional Autoencoder (CAE
Park, Jun-Seo, Jo, Hyeon-Choel, Cho, In-Je, Seo, Jae-Yong, Yoo, Seong-Sik
Challenges in Predicting Fatigue Damage during Agricultural Operations: Insights from Machine Learning Models02-19-03-00176/10/2026
Agricultural vehicles operating in rough environments experience increased fatigue damage accumulation, which may decrease machine safety and reliability. Autonomous agricultural machines offer an opportunity to incorporate fatigue damage considerations into path planning. This work investigates whether machine learning can predict fatigue damage to a tractor chassis using light detection and ranging (LiDAR)-based terrain features, vehicle speed, and rotational vehicle state data (e.g., triaxial angle, angular velocity, and angular acceleration). Fatigue damage was estimated using the Rupp filter and the Durability Transfer Concept. Following poor predictive performance of the machine learning models, an exploratory analysis of damage histograms, dominant frequency, and acceleration magnitude was performed. Results indicated that most estimated fatigue damage occurred in the 0–2 Hz band, which coincides with the frequency range of terrain-induced acceleration. On-road driving led to
Govers, Megan Emily, Hamilton-Wright, Andrew, Hassan, Marwan, Oliver, Michele L.
Can Large Language Models Enable Sustainable Driving Behavior? A Study on Large Language Model-Based Eco-Driving13-07-02-00066/9/2026
Large language models (LLMs) have shown remarkable capabilities for perceiving driving environments and making interpretable, logical decisions for autonomous driving. However, their potential for more comprehensive driving strategies, especially concerning energy efficiency, remains underexplored. Most existing studies primarily focus on driving safety, which may inadvertently increase energy consumption. To address this issue, this study explores the use of LLMs as high-level controllers to jointly optimize driving safety and energy efficiency. A textual prompt is designed for the LLM, incorporating few-shot examples that describe scenarios, states, and actions. The LLM processes the scenario and state prompts describing the surrounding traffic environment. It generates a high-level control signal, which is then translated into low-level vehicle motion commands in a high-fidelity traffic simulator with realistic physics, vehicle dynamics, road slopes, and network topology
Wang, Haoyu, Li, Zhenning, Wang, Siying, Zhou, Zijing, Zhang, Xiang, Yang, Zhifeng, Ou, Shiqi (Shawn), Qi, Hao
Rapid Fuel Distribution Prediction in RCCI Marine Engines Using CFD-Informed, Active Learning Framework2026-37-00176/9/2026
Accurate prediction of in-cylinder fuel distribution (FD) is fundamental to reduced-order combustion modeling and emissions prediction yet remains computationally prohibitive with high-fidelity CFD alone. This work develops a CFD-informed machine-learning surrogate for spatial FD in a large-bore diesel engine, based on a Wärtsilä W20 injector and representative engine conditions. A fully coupled injector–spray–engine CFD framework under engine-like RCCI inert conditions determines the needle-lift profile and resolves the combined effects of injector geometry, needle dynamics, and operating conditions on in-cylinder flow, capturing physical phenomena not reproducible by isolated free-spray simulations. A high-fidelity database is generated using Latin Hypercube Sampling, from which FD is extracted at 15 CAD before top dead center within an annular multi-zone (MZ) representation consistent with reduced-order combustion models. A multi-output Random Forest (RF) surrogate, augmented with
Moradi, Jamshid, Salahi, Mahdi, Heidarabadi, Shadab, Andwari, Amin, Konno, Juho, Wik, Christer, Mikulski, Maciej
SAE TOMORROW TODAY - The Battery Isn't the Problem--Execution Is135716/8/2026
Batteries don't fail because of chemistry. They fail because of execution. Qnovo provides advanced software to optimize lithium-ion battery health and safety, proving that competitive advantage isn't just about the battery itself, but how intelligently you manage it. By leveraging sophisticated battery models enhanced with AI and machine learning, the company delivers real-time insights into battery performance and operation. Listen in as we sit down with Nadim Maluf, Co-founder and CEO, to discuss how software has become the soul of modern batteries. From extending smartphone battery warranties to enabling faster, safer EV charging, you'll discover why the future of electrification isn't about new materials ... it's about intelligence, scale, and trust. We'd love to hear from you. Share your comments, questions and ideas for future topics and guests to podcast@sae.org. Don't forget to take a moment to follow SAE Tomorrow Today--a podcast where we discuss emerging technology and trends
Patterson, Lori
Weight Estimation of Novel Aircraft Configurations in Early Design Phase2026-26-07736/1/2026
Initial weight estimation from Top Level Aircraft Requirements (TLAR) is a critical first step in aircraft design, yet existing empirical methods are inadequate for novel configurations such as those using Liquid Hydrogen (LH2) or Sustainable Aviation Fuels (SAF). This paper presents a hybrid methodology for top-level weight estimation of such unconventional aircraft. The approach is based on modifying a conventional baseline aircraft, integrating a new statistical model with component-specific weight estimations. A multivariate regression model to estimate the empty weight fraction (We/W0) was developed from a dataset of 44 conventional aircraft, yielding an R-squared value of 0.833. This statistical model was integrated with physics-based models for novel components, including cryogenic fuel tanks and fuel systems. The methodology accounts for iterative changes to fuselage structure and parasitic drag. Four configurations were analyzed: fuel types being Jet A1, SAF, LH2 with aft
Goyal, Tushar
Integrating Human factors and Predictive Modeling for Enhanced Aviation Safety: A Data - Driven Approach to Risk Mitigation and System Design2026-26-07956/1/2026
This study presents a data-driven approach for strengthening aviation safety by integrating human factors assessment with modern predictive modeling techniques. The work focuses on understanding how human performance, operational conditions, and system-level interactions collectively influence safety risk, and how these interactions can be quantified to support improved design and decision-making. Unlike previous studies that address human factors or predictive modeling in isolation, this research offers a unified framework that links causal human factors indicators with statistical modeling, feature extraction, and machine learning based risk estimation. The novelty of this work lies in the structured pipeline that transforms raw categorical and narrative human factors information into measurable predictors that can be analyzed using structural modeling and machine learning. The methodology includes data preparation, dimensionality reduction, latent pattern discovery, dependence
Valiyaparambil, Praveen
Drilling Studies on Sustainable Lightweight Hybrid Polymer Matrix High-Performance Nanocomposites2026-26-07326/1/2026
Worldwide, engineers are exploring the possibility of using polymer composites in their quest for lightweight materials. In this study, injection moulding was used to develop a biodegradable polymer PLA composite containing 20 wt.% vetiver fibers (VFs) and 2 wt.% nano-silica (nSiO2) obtained from pearl millet, which is sustainable. Materials need machining as secondary operation that required joining. Desirability analysis was used to examine and optimize machining (drilling) studies that were designed with Taguchi's design (L9 orthogonal array). Surface roughness (SR) and delamination factor (Fd) were taken as outputs, while spindle speed (SS), feed rate (FR), and drill diameter (DD) were the inputs. Drilling studies were performed on a single vertical machining center (VMC). ANOVA identifies that the FR had the most decisive influence on SR (F=559.24, p=0.001785), followed by DD and SS. FR is the dominant contributor to Fd (F=379, p=0.00263), followed by SS and DD. At low SS and high
Senthilkumar, N.
AI-Led Sustainability Strategy: Driving Product Value from Birth to Disposal2026-26-07816/1/2026
Aerospace products operate within highly complex, safety-critical environments and endure extended lifecycles, often spanning decades. Sustaining their operational value requires rigorous management of Safety, Reliability, and Availability (SRA), while global Environmental, Social, and Governance (ESG) mandates demand parallel progress toward sustainability goals. This paper introduces an AI-driven strategy that integrates these dual imperatives—Sustenance Management and Sustainability Management—within a unified Product Lifecycle (PLC) framework. The proposed approach leverages Artificial Intelligence across five PLC phases: Generative Design, Detailed Design & Verification, Manufacturing & Industrialization, Operations & Maintenance, and End-of-Life Circularity. Anchored by a certified Digital Thread, this framework ensures seamless, auditable data flow from concept to disposal. Using Life-Limiting Parts (LLPs)—such as high-stress turbine discs—as a case study, the paper demonstrates
Srinivasan, Karthik, G.V.V., Ravi Kumar, Vaderahobli, Devaraja Holla, Bhate, Ujwal, Veluri, Sastry
Hybrid Digital Twin for Aero-Engine Bearing Prognostics: A Physics-Informed Machine Learning Approach to RUL Prediction2026-26-07186/1/2026
Unscheduled maintenance due to the failure of critical components, such as aero-engine rolling element bearings, is a leading cause of costly Aircraft-on-Ground (AOG) events; consequently, current time-based maintenance practices are inefficient and prone to risk. This paper develops a resource-efficient Hybrid Digital Twin (HDT) model for an engine bearing, focusing on the dynamic prediction of spall growth due to Rolling Contact Fatigue (RCF), thereby enabling a condition-based maintenance paradigm. The HDT architecture integrates two core models: (1) a physics-informed model that uses established life and fatigue theory to define initial degradation thresholds, and (2) a data-driven Recurrent Neural Network (RNN), specifically a Long Short-Term Memory (LSTM) network, for dynamic degradation rate modeling. The methodology utilizes a Monte Carlo simulation coupled with RCF progression equations to generate a large, high-fidelity synthetic run-to-failure dataset under varying
Mohamed, Abbas
AI Based Energy Sustained Series Hybrid Aircraft2026-26-07766/1/2026
This paper investigates the energy consumption characteristics of series hybrid aircraft with a focus on comparing conventional energy management approaches against an AI-powered optimization framework. The study comprehensively models the energy demands of a series hybrid aircraft across all major flight phases, including Idle & Ground Operations, Taxi, Takeoff, Climb, Cruise, Descent, Approach, Landing, and Rollout & Taxi. For each phase, detailed mathematical formulations are developed to capture power requirements and energy flow, incorporating real-time operational parameters to enhance the accuracy of the energy consumption estimations measured in kilowatt-hours (kWh). The AI-based optimization leverages advanced control strategies, specifically Model Predictive Control (MPC) and Reinforcement Learning (RL) algorithms, to dynamically manage the aircraft’s energy systems. MPC is employed to predict and optimize future energy usage by solving constrained optimization problems over
Kanchagar, Amogha
Predictive Model and Analysis of the Resultant Force in the Dry Machining of Mg-4Zn/Si 3 N 4 Nanocomposites2026-26-07626/1/2026
The development of lightweight materials for use in aerospace and automotive applications is extremely significant. Magnesium (Mg)-based alloys and composites are good candidate materials from the perspective of low density, good specific strength, and abundance. The Mg-4Zn alloy is one such alloy, which is a lightweight, biocompatible, and eco-friendly Mg-based alloy. In spite of these advantages, there is a strong need and scope to improve its wear resistance and mechanical properties. Mg-4Zn nanocomposites with Si3N4 reinforcements (a biocompatible bioceramic) are hypothesized to possess superior properties. Microstructural analysis of the vacuum stir-cast nanocomposites confirms grain refinement and a consequent increase in microhardness with an increase in Si3N4 reinforcement wt.%. The addition of Si3N4 reinforcement to improve the properties of the Mg-4Zn alloy could introduce challenges in machining. To make products from the nanocomposites, machining them with minimal
N, Anand, Shaju, Tony M, G, Nagamalleswara Rao, D, Bijulal, K, Jayaprakash Reddy, K, Vijayan, Chaman, Joji J
Automated Flight Manoeuver Recognition for Intelligent Flight Simulators2026-26-07136/1/2026
The rapid growth in the number of aircraft and pilots emphasises the need for an AI-enabled training framework that can offer precise, automated examination of flight manoeuvres. This will be useful in optimising the pilot's training efficiency and minimising iterations of the conduct of flight manoeuvres, thereby reducing the training time of the pilot for a flight. A general framework is developed that can be used for all kinds of flight phases and aircraft types. A pre-trained machine learning model is designed using a supervised learning technique, Random Forest, to recognise different manoeuvres. Various statistical parameters, such as mean, standard deviation, kurtosis, skewness, etc., of several flight parameters were used as the input features to train the Random Forest classifier. In the present work, the classifier is trained using several actual flight test data manoeuvres, and is also supplemented with simulated manoeuvres. The achieved gross accuracy for manoeuvre
Sahu, Akash, C, Poornima, C, Aravindh, Kaliyari, Dushyant, TK, Khadeeja Nusrath
Adaptive HVAC Strategies for Enhanced Cabin Thermal Comfort, Air Quality and Energy Efficiency2026-26-07916/1/2026
Passenger comfort within vehicles and aerospace cabins relies on finely tuned management of temperature, air quality, and energy use. This paper proposes an integrated HVAC framework that combines zonal climate control, intelligent airflow distribution, and real-time sensor data to maintain thermal balance across different cabin zones. Leveraging predictive thermal load modelling and machine learning, the system anticipates environmental changes—such as sudden shifts in external temperature or passenger load—and proactively adjusts heating and cooling outputs. Simultaneously, air quality is enhanced through a multistage filtration system, active air purification technologies, and dynamic CO₂ concentration monitoring. Comfort assessment integrates PMV (Predicted Mean Vote) and PPD (Predicted Percentage Dissatisfied) indices to adapting environmental conditions. Simulations and early-stage prototypes improve energy savings and improve occupant comfort and air quality. The proposed HVAC
Mudavath, Lehitha Sai, Patil, Ashish, Saha, Sudipta
Aircraft Cabin Diagnosis – Inspection of the Aircraft Cabins and Seats Using AI2026-26-07896/1/2026
Aircraft interior defects, including seat structural damage, cushion degradation, liquid contamination, and foreign object presence, contribute to increased maintenance burden, extended ground time, and operational inefficiencies. Current inspection practices rely predominantly on manual visual checks, which are time-intensive and limited in detecting concealed anomalies. This paper presents a non-contact, AI-enabled inspection framework integrating millimeter-wave (mmWave) radar sensing with high-definition optical imaging for automated aircraft seat condition assessment. The proposed system captures interior scans when the aircraft is unoccupied and compares them against a digitally established baseline reference obtained under certified, defect-free conditions. Data fusion and machine learning algorithms analyze deviations to identify surface and subsurface defects at seat-level resolution and generate zone-based maintenance maps. The primary technical contribution lies in combining
Nagoal, Chandrasekhar Reddy, Prathipati, Krishna Chaitanya, Kandukuri, Ravindra
Data-Driven Air Traffic Management: Forecasting Demand, Modeling Delay Propagation, and Demand Capacity Balancing2026-26-07726/1/2026
Air Traffic Management (ATM) must be familiar with the exact Aircraft Take-off Weights (ATOWs) of airplanes to make the most use of runways, maintain safety margins high, and keep utilization and resources in balance. This paper aims to present a dependable ATOW forecasting methodology that can assist the air transport industry in enhancing operational decision-making. This research used datasets acquired from the EUROCONTROL Performance Review Commission (PRC) 2024 Aircraft Take-Off Weight Estimation dataset featuring 527,000 flights over Europe containing aircraft details, air trips and flight conditions. Technique comprises structured data input, inspection of missing data, timestamp aggregation to identify demand cycles over time, and domain-specific feature engineering using distance_per_minute, block_minutes, taxiout_ratio, and a strong wake turbulence metric The two supervised learning models used were Linear Regression (LR) for understanding and XGBoost for performance
Senthilkumar, N., S, Gopalakrishnan, Gopinath, S
Zero-Shot Classification of Additive Manufacturing Anomalies2026-26-07216/1/2026
Additive Manufacturing (AM) process involves building part layer by layer. Some of the AM processes ( Laser and Electron beam based) generate a melt pool during printing process. This melt pool can be captured periodically during AM process using special optical arrangements. These images capture high intensity melted zone, heat affected zone, splattered molten metal particles and overall shape of the melt pool. These images carry similar characteristics for good AM processes within a range. When there is an anomaly the above said characteristics of the melt pool changes, for example a low intensity melted zone signifies low energy condition which can lead to defects like balling etc. Hence the captured image at this condition appears significantly different from other images. The common defects which can be detected by analyzing melt pool images are porosity, spatter, lack of fusion, cracks, balling and keyhole instability. There are many machine learning methods available to quantify
Kuppusamy, Balasundar
PMHS Sled Testing of Reclined Small Female Occupants: Pelvic Dynamics and Injury Evaluation2026-22-00015/28/2026
As automated vehicle technologies enable increased seat recline angles during travel, understanding the biomechanics of injury under these novel occupant postures becomes imperative. This study evaluated the pelvis injury response and associated kinematics of reclined small female post-mortem human surrogates (PMHS) subjected to frontal sled tests across three restraint configurations. Each configuration varied in seat stiffness and the presence of a knee bolster to assess their influence on pelvic dynamics and submarining risk. Nine PMHS tests were conducted using a consistent reclined posture (38° thorax, 75–80° pelvis angle) and production restraint systems. Submarining probability was estimated using a validated logistic regression referenced from previous study. Distinct pelvic kinematics, fracture patterns, and associated injury mechanisms emerged across the test configurations in the current dataset. Configuration 1, featuring a stiffer seat without a knee bolster, exhibited
Somasundaram, Karthik, Driesslein, Klaus, Pintar, Frank A.
Energy Saving and Personalized Thermal Comfort Control Based on Reinforcement Learning and Decision Tree2026-99-17145/22/2026
Indoor thermal comfort is closely related to people’s health and work efficiency. Control systems typically consume a large amount of energy to maintain a comfortable thermal environment. Currently, reinforcement learning is widely applied to optimize thermal comfort control systems. However, existing research mainly adopts universal thermal comfort evaluation models that aim to satisfy the majority of people, which makes it difficult to quickly and accurately reflect the specific thermal comfort needs of individuals. As a result, the hot environment is neither comfortable nor energy-efficient in practical use. Therefore, this paper proposes an energy-saving personalized thermal comfort control method based on decision trees and reinforcement learning. First, decision tree learning is used to obtain an individual thermal comfort evaluation model from a small amount of historical data. Then, this individual comfort model is combined with energy consumption to form a reward function
Li, Xianying
DDPG-Based Energy-Optimal Torque Split Strategy for Dual-Motor Electric Vehicles10-10-04-00335/22/2026
This study presents a torque distribution strategy for dual-motor electric vehicles utilizing a Deep Deterministic Policy Gradient reinforcement learning algorithm designed to optimize energy consumption. By using a simplified architecture and replicable reward functions, the proposed agents rely exclusively on standard CAN bus signals, commanded longitudinal force, and the motors’ velocities, eliminating the need for specialized sensors or complex plant models. Two reinforcement agents are trained using two different reward functions: power-based and State of Charge-based. These agents are validated through high-fidelity CarSim–Simulink co-simulations across soft, medium, and severe acceleration scenarios, in which they demonstrate superior performance to traditional adaptive methods. In the most demanding scenario, a typical adaptive strategy achieves an additional 7.8% of power consumption and 85% of optimal energy recovery, while the proposed reinforcement learning strategies reach
Meléndez-Useros, Miguel, Viadero-Monasterio, Fernando, López-Boada, María Jesús, López-Boada, Beatriz
Hierarchical Reinforcement Learning For Task Generation of Transmission Tower Inspection Using Point Clouds2026-99-07545/15/2026
Unmanned Aerial Vehicles (UAVs) are widely used for inspecting transmission towers. However, traditional waypoint planning relies heavily on manual experience. This leads to low efficiency, incomplete coverage, and a lack of standardization. Facing these problems, this paper proposes an intelligent generation method based on Hierarchical Reinforcement Learning (HRL). This method achieves end-to-end automation, converting raw point cloud data directly into an optimal set of waypoints. Preprocess and grid the point cloud data to build a model of the coverage area. Then design a hierarchical framework to break down the complex planning task. This framework divides the task into high-level waypoint selection and low-level pose optimization. Specifically, the high-level part uses a Deep Q-Network (DQN) to learn the best sequence of waypoints. The low-level part uses Q-learning tables to optimize the pitch and yaw angles for each point. Meanwhile, design a reward function to maximize
Cui, Shicheng, Lin, Shizhong, Shao, Zhan, Chen, Ruiduan, Li, Xingyu, Luo, He
A Deep Learning Framework for Fast Prediction of Automotive Drag and Pressure Fields2026-99-07485/15/2026
Computational fluid dynamics (CFD) is crucial for automotive design, requiring analysis of 3D point clouds to investigate how vehicle geometry affects pressure fields and drag. Running CFD on high-resolution 3D geometry quickly becomes computationally heavy, and many solvers slow down noticeably as the geometric detail increases. We therefore introduce a dual-task deep learning framework, named AeroFormer, that predicts aerodynamic quantities directly from the vehicle’s surface geometry and avoids the need for full CFD simulations. The model is organized into two parts. One branch, AeroFormer-Cd, predicts the overall drag coefficient (Cd), while the other, AeroFormer-Press, reconstructs the pressure distribution over the vehicle’s surface. Both branches rely on a shared curvature-guided adaptive sampling process and a physics-aware attention encoding module, which enable the network to emphasize fine geometric details in aerodynamically sensitive regions such as the front bumper, A
Yan, Shengmao, Deng, Shisong, Jiang, Yanzhen, Jin, Xinyu, Cai, Zhengyang
A Reinforcement Learning Trajectory Planning Model with Kalman Filter-Based Dynamic Tracking for Six-Axis Robotic Manipulators2026-99-07385/15/2026
Robot Arm Tracking Control refers to the control of robot end effectors following a prescribed trajectory as their movement in robotic systems. The work presents a combination of Kalman Filter Based Dynamic System Tracking with Reinforcement Learning Based Trajectory Planning. These two aspects of tracking and planning help the robotic manipulator dynamically track a target that is located on an arbitrary moving path. In particular, by using Kalman filtering to estimate the position of a moving target and to compensate for sensor noise and sparse sampling, we take high-precision estimation values of each point’s coordinates along the target trajectory as a reliable basis to build a policy network using reinforcement learning. Based on it, the robot manipulator could produce effective motion planning under its own dynamic capabilities and physical constraint limit. Comprehensive simulation results illustrate advantages of the new algorithm against the classical control method, confirm
Yu, Jingze, Wang, Yujia, Li, Junshen, Chen, Cong, Xu, Peng
Time Series Causal Discovery Based on Data Augmentation and Causal Validation2026-99-07035/15/2026
Causal discovery within time series is crucial for revealing the actual causal mechanisms in dynamic systems, and it has major impacts in various fields like economics, healthcare, and climate science. Even though it’s important, accurately figuring out causal relationships from observational temporal data is still quite a difficult task. Traditional Granger causality based methods are often limited by noise sensitivity, large amount of data, and the inability to distinguish between real causality and false correlation caused by hidden factors. In order to solve these problems, this paper presents CausalAugVeri, which is a new algorithm that cleverly mixes data augmentation with causal verification to make causal discovery more solid and precise. This work has three main points: First, we carefully check that using convolutional data augmentation techniques can greatly improve how well time series predictions work, giving a steadier base for detecting Granger causality. Second, the
Yang, Jing, Chen, Xiaotao, Qin, Xuanli, Xu, Xianjun, Hu, Zhangxiang
Feature Reduction for Scenario-Based Safety Assessment of Automated Driving Systems09-14-01-00185/4/2026
Vehicles equipped with an Automated Driving System (ADS) have the potential to significantly reduce road collisions. To enable widespread adoption of ADSs, rigorous safety assessment is essential. Valuable insights for ADS safety validation can be gained by simulating scenarios across a broad range of feature variations. A common challenge in simulating these scenarios is known as the curse of dimensionality, where increasing the number of scenario features requires a near-infinite number of simulations to cover all variations. This issue of complexity presents a need for reducing scenario features. Most related work focuses on identifying important scenario features, while few evaluate how reducing these features impacts ADS failure estimation. The present study aims to address this gap by employing a wide range of feature reduction methods and assessing their effect on ADS failure estimation. Previous research generated datasets for three distinct scenario categories by performing
Lankhorst, Bram, de Gelder, Erwin, Janssen, Christian P., Scholich, Andre
Rear-Impact Crash Injury Patterns and Predictors: Insights from Contemporary CISS and CIREN Field Data09-14-01-00295/4/2026
The objective of this study was to investigate occupant injury patterns and predictors in rear-impact crashes using recent US field data. Cases were queried from the Crash Investigation Sampling System (CISS, 2017–2023) and the Crash Injury Research and Engineering Network (CIREN, 2017–2024), yielding 1923 front-row outboard occupants from 1533 crashes. Crash documentation and vehicle photographs were manually reviewed to classify seatback deformation magnitude and secondary impact severity. Multivariable logistic regression models estimated associations between occupant, vehicle, and crash characteristics and Abbreviated Injury Scale (AIS) ≥ 2 and AIS ≥ 3 injury outcomes across body regions. Sensitivity analyses included CISS-only, weighted, single-event, and interaction models. Thoracic injuries were further subdivided into skeletal and cardiopulmonary categories. Findings reflect associations within the pooled CISS + CIREN analytic sample rather than nationally representative injury
Lockerby, Jack, Rudd, Rodney
Effects of Vehicle Speed, Category, and Front-End Design on Full-Body Injury Risks of a Midsized Male Pedestrian09-14-01-00175/4/2026
This study investigated how vehicle front-end geometry, impact speed, and vehicle category influence injury risk to a midsize male pedestrian. Eighty-one generic vehicle (GV) models representing sedans, sport utility vehicles (SUVs), pickup trucks, and minivans sold in the United States were developed by morphing three base models using an automated pipeline. Front-end parameters that were varied included ground clearance (GC), bumper height (BH), hood leading-edge (HLE) height, hood length (HL), bumper lead angle (BLA), hood angle (HA), and windshield angle (WSA). Each vehicle impacted the Global Human Body Models Consortium 50th percentile male simplified pedestrian (GHBMC M50-PS) model at 30, 40, and 50 kph, totaling 243 simulations. Boundary conditions followed the European New Car Assessment Program (Euro NCAP) pedestrian test protocol. Thirty-five injury metrics were extracted across the head, neck, thorax, abdomen, pelvis, and lower extremities. Linear mixed-effects regression
Poveda, Luis, Miller, Logan E., Edwards, Colin C., Pollock, Madeline, Armstrong, William M., Hsu, Fang-Chi, Gayzik, Scott F., Weaver, Ashley A., Stitzel, Joel D., Devane, Karan S.
Machine Learning–Driven Safety Analysis of Automated Vehicles in Simulated Mixed Traffic Environments09-14-01-00215/4/2026
Automated Vehicles (AV) pose new challenges in road safety, multimodal interaction, and urban planning, requiring a holistic approach that prioritizes sustainability and protects all road users. The KASSA.AST project addresses this by deploying and evaluating an automated shuttle in southern Austria on three routes. The study area is a Park & Ride zone near a train station, enabling seamless transfers and higher transit use. To assess the safety impacts of the automated shuttle, four Mobility Observation Boxes (MOBs) were deployed. These AI-based systems detect and classify road users, track their trajectories and geospatial coordinates, and identify safety-critical events via Surrogate Safety Measures (SSMs). Over 10 days, a trajectory dataset captured interactions among vehicles and the shuttle. The resulting real-world dataset is a core contribution. This dataset underpins microscopic behavior modeling. Trajectory pairs yield car-following and interaction metrics (relative distance
Losada Arias, Ángel, Rosenkranz, Paul, Hula, Andreas, Aleksa, Michael, Saleh, Peter, Erdelean, Isabela
A Review of Data-Driven Approaches and Datasets for Battery State of Health Estimation14-15-02-00094/22/2026
This article surveys the most recent data-driven methods of lithium-ion (Li-ion) battery state of health (SOH) estimation methods and dataset resources utilized in electrified vehicles (EV) and their potential adoption for automotive battery management systems. These include regression-based models, ensemble learners, deep neural networks, and physics-informed hybrid methods. The review describes estimation methods found in articles published between 2023 and 2025, and investigates their differences in terms of estimation accuracy, data requirement, interpretability, and real-time deployment ability. The article traverses the dataset space, focusing on laboratory aging datasets, vehicle field–based datasets, telematics-derived records, and synthetic or augmented datasets, to underline that model performance in the estimation of SOH cannot be disentangled from the quality of the data, the operating coverage, and the transfer conditions. Apart from the model design, this work reviews the
Nyachionjeka, Kumbirayi, Bayoumi, Ehab H.E.
The ML FMEA in Action: Lessons from Applications of Machine Learning Safety2026-01-00794/7/2026
Introducing machine learning (ML) into safety-critical systems presents a fundamental challenge, as traditional safety analysis techniques often struggle to capture the dynamic, data-driven, and non-deterministic behavior of learning-enabled components. To address this gap, the Machine Learning Failure Mode and Effects Analysis (ML FMEA) methodology was developed as an open-source framework tailored to ML-specific risks. This paper reports on the maturation of ML FMEA from an initial conceptual framework to a proven, practice-driven methodology. We make four primary contributions. First, we extend the ML FMEA pipeline with two new stages: a “Step Zero” for problem definition and system-level hazard analysis, and a “Step 5” for constructing ground truth or reward signals. Autonomous vehicle and humanoid robot applications are presented to illustrate the practical application and safety benefits of these additions. Second, we introduce tailored Severity, Occurrence, and Detection
Schmitt, Paul, Shinde, Chaitanya, Diemert, Simon, Pennar, Krzysztof, Seifert, Bodo, Poh, Justin, Lopez, Jerry, Mannan, Fahim, Mohammed, Majed, Chalana, Akshay, Wadhvana, Neil, Wagner, Michael
A Dataset for Visual Classification of Flat Tires2026-01-01524/7/2026
Flat tires represent a common yet serious issue in vehicle safety, leading to compromised control, increased braking distance, and potential rim or structural damage when undetected. Conventional tire pressure monitoring systems (TPMS) rely on embedded sensors that can fail, incur high replacement costs, and are not always equipped in older or low-cost vehicles. To address these limitations, this study presents a comprehensive visual dataset for flat-tire classification using computer vision and machine learning techniques. The dataset comprises 600 labeled images—300 flat-tire and 300 non-flat-tire samples—collected from diverse vehicle types, lighting conditions, and viewpoints. This dataset is designed to support the training and benchmarking of lightweight edge-AI models suitable for real-time deployment on embedded platforms. A set of supervised learning models were evaluated. Results demonstrate that visual-based classification provides a cost-effective and scalable pathway
Gunasekaran, Aswin, Govilesh, Vidarshana, Challa, Karthikeya, Maxim, Bruce, Shen, Jie
A Latent Space Framework for Modeling Transient Engine Emissions Using Joint Embedding Predictive Architectures2026-01-04234/7/2026
Accurately modeling and controlling vehicle exhaust emissions, particularly during highly transient events such as rapid acceleration, is crucial for meeting stringent environmental regulations and optimizing modern powertrain systems. While conventional data-driven modeling methods, such as Multilayer Perceptrons (MLPs) and Long Short-Term Memory (LSTM) networks, have improved upon earlier phenomenological or physics-based models, they often struggle to capture the complex nonlinear dynamics of emission formation. These monolithic architectures attempt to learn from all available data, which increases their sensitivity to dataset variability. They often require increasingly deep and complex architectures to improve performance, thereby limiting their practical utility. This paper introduces a novel approach that overcomes these limitations by modeling emission dynamics in a structured latent space. Using a rich dataset combining real-world driving data from a Portable Emission
Sundaram, Ganesh, Gehra, Tobias, Ulmen, Jonas, Heubaum, Mirjan, Görges, Daniel, Günthner, Michael
Multi-Physics Topology Optimization and Neural Network–Based Design of Experiments for Lightweight Automotive Cast Components2026-01-05104/7/2026
This paper presents a hybrid optimization framework that integrates Multi-Physics Topology Optimization (MPTO) with a Neural Network–surrogated Design of Experiments (NN-DOE) to enable lightweight structural design while satisfying crashworthiness, durability, and noise, vibration, and harshness (NVH) requirements under practical casting and packaging constraints. In the proposed MPTO formulation, crash and durability performances are incorporated through equivalent static compliance measures, while NVH performance is assessed using a frequency-domain dynamic stiffness metric, allowing consistent evaluation of trade-offs among competing design requirements. The framework is first demonstrated using a mass-produced passenger-car lower control arm (LCA) as a benchmark component. In this application, MPTO achieves weight reduction under multi-physics objectives by removing non-load-bearing material. Results show that single-discipline optimization produces unbalanced topologies, while
Kim, Hyosig, Senkowski, Andres, Gona, Kiran, Saroha, Lalit, Boraiah, Mahesh
Items per page:
1 – 50 of 1496