Korea Automotive Technology Institute

Robotic milling with a KUKA KR210 R2700-2 offers flexibility but suffers from chatter, corner rounding, and variable surface finish. This study proposes an AI-assisted robotic milling framework using a CNN–BiLSTM architecture with transfer learning to improve machining precision and surface quality. The model processes real-time vibration, force, and acoustic signals to predict tool wear and adaptively adjust cutting parameters. Compared with manual milling, the proposed approach achieves smoother surface finish, reduced chatter, and more stable feed modulation. An optional online guard adjusts feed override using vibration and temperature signals. In benchmarking tests on aluminum, polymer, and wood benchmarks, the method improves surface quality and maintains or reduces cycle time without modifying upstream CAM. Experimental results on aluminum and polymer show superior performance with MAE = 1.50mm, RMSE = 0.95mm, and R<sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup> = 0.99, confirming the model’s high accuracy and efficiency for intelligent robotic milling control.

This study analyzes the lifetime characteristics and degradation behavior of lithium-ion batteries under increasing charge–discharge cycles. The experiment focused on RE (Real Part), IM (Imaginary Part), and DCIR Degradation% (Direct Current Internal Resistance Degradation). The RE increased from 0.0023 Ω at the initial state to 0.00293 Ω after 1200 cycles, representing a 28% rise, with a sharp acceleration after 400 cycles due to interfacial resistance buildup and electrolyte decomposition. The IM shifted from negative to positive values, indicating delayed electrochemical reactions and enhanced inductive behavior. A pronounced transition occurred between 400 and 800 cycles, confirming this range as a critical phase of performance degradation. Correlation analysis between SoH (State of Health) and DCIR Degradation% showed that while SoH decreased slightly from 100% to 87.3%, DCIR Degradation% increased significantly to 137.8%, indicating that internal resistance growth is the dominant cause of aging. When SoH falls below 70%, the battery reaches its effective end-of-life, accompanied by severe heat generation and power loss. In conclusion, the combined analysis of RE, IM, and DCIR Degradation% demonstrates that accumulated internal resistance is the key factor determining battery lifetime. Stabilizing the SEI layer, reinforcing electrode structures, and improving electrolyte stability are essential strategies for extending battery durability.

This study presents a vision-based approach for estimating surface roughness in automotive press dies. Four feature extraction algorithms are evaluated: Grey Scale Average (GSA), FFT, Sharpness, and Texture Homogeneity. Grayscale images are collected from die surfaces, and each image is paired with a surface roughness value measured using a contact-type profilometer. Single scalar features are extracted by each algorithm and used to train regression models, including linear, polynomial, exponential, and logarithmic functions. Model performance is assessed using R-squared, root mean square error, and relative error. GSA and FFT models show the highest accuracy, with R-squared values near 0.66. These models capture global brightness and frequency-domain patterns, which exhibit stronger correlations with surface variation than local edge sharpness or texture features. The results suggest that global image characteristics are effective for surface roughness estimation. This method supports non-contact surface quality monitoring in die manufacturing, and future research may improve its accuracy and robustness through advanced modeling techniques.

Owing to the rapid popularization of electric vehicles (EVs), high-voltage power-converter reliability is gaining importance. DC-DC converters are components of EV power-conversion systems that provide and control the voltage required by electronic devices that use high-voltage battery sources. Reliability issues are caused by electromagnetic interference (EMI) between electronic devices, while driver safety concerns are caused by the use of high-capacity batteries. Design topologies of high-voltage EV DC-DC converters include the LLC resonant topology, which enables zero-voltage switching (ZVS) and zero-current switching to minimize electrical losses in high-speed switching environments. This study analyzed the EMI-conducted emission (CE) noise under ZVS and proposed optimal design factors for noise reduction.

This paper presents the method for reducing AC copper loss by adding the optimal air gap at commercial transformer on the CLLC converter. In order to figure out the optimal air gap, FEM analysis was progressed and the transformer was implemented. To verify the proposed method, the experiments were performed with 11kW CLLC converter and the AC copper loss of transformer was reduced by about 10W compare with conventional transformer model.

Abstract This study carried out experimental rolling contact fatigue (RCF) tests on two types of materials: commercially available Cr-Mo alloy steel and 1.55% Ni-doped Cr-Mo alloy steel, using a twin-disc test rig. Micropitting was identified as the most prominent damage feature on the rolled surfaces of the disc samples during the gears' RCF tests. To investigate gear surface characteristics and wear behavior, RCF tests were conducted on both material types. The twin-disc test was designed to replicate asperity contact on mating gear flank surfaces, using low-speed cylindrical disc specimens and high-speed crowned disc specimens, ensuring a minimum effective contact area of 8.5 mm. After testing, surface morphology and topographical changes were examined using Scanning Electron Microscopy (SEM). The results revealed that 1.55% Ni-doped Cr-Mo alloy steel demonstrated superior resistance to contact fatigue failure compared to the commercially available Cr-Mo alloy steel, as evidenced by a lower micropitted area ratio, shallower pitted depth, and fewer pits. Specifically, the Ni-doped alloy exhibited a 5.66% lower micropitting area ratio under the same test conditions. Furthermore, prior studies support these findings, showing that the 1.55% Ni-doped steel performs better than the 2% Ni-doped alloys. Therefore, 1.55% Ni-doped Cr-Mo alloy steel is recommended for transmission gears to minimize RCF damage.

Durante o processo de desenvolvimento de um veículo, os ensaios destrutivos de colisão de veículo completo (crash tests) representam um investimento significativo para os fabricantes, principalmente quando são necessários vários loops de ensaio. Para reduzir esses custos substanciais na construção de protótipos, foi proposta uma solução onde os testes de poste lateral em escala real podem ser simplificados para testes de trenó de subsistema, exigindo menos recursos em termos de carrocerias e, ao mesmo tempo, permitindo reproduzir condições de testes de impacto de poste lateral, onde a desaceleração do veículo e a intrusão de poste são aplicadas de forma controlada para reproduzir a cinemática da colisão. Além disso, em protocolos globais de segurança passiva, um dos casos de carga mais exigentes e complexos para o desenvolvimento de sistemas de contenção lateral é o impacto lateral a 32 km/h contra um poste rígido com o manequim WorldSID na posição do motorista, que representa um veículo colidindo lateralmente com objetos rígidos na beira da estrada, como árvores ou postes. Este artigo explora o desenvolvimento desta configuração de teste de subsistema para este caso de carga, que pode ser aplicado para o desenvolvimento do sistema de contenção.

The article covers how three-dimensional packing affects the shipping cost under dimensional weight pricing, and the algorithmic plus engineering solutions through which packing geometry is converted into a direct economic effect.Its objective is to make a 3D packing algorithm e-commerce ready, compare constructively with metaheuristic approaches, estimate ML and DRL roles in online formulations, justify integration of stability validators and compatibility rules, and sketch an end-to-end scheme for embedding into a WMS with early rating of candidates.The relevance is driven by rising parcel volumes and carrier practices that charge for the maximum of actual and dimensional weight, which turns every extra centimeter into an overpayment and raises the bar for millisecond decisions on the fulfillment line.The novelty lies in formulating the objective as a forecasted cost that accounts for carrier divisors and zonal rates, and in describing a hybrid cascade in which fast heuristics of the EB-AFIT class and their implementations in 3DContainerPacking and Sharp3DBinPacking provide a baseline plan, while ALNS and GA improvers, including a GAN+GA hybrid, as well as DRL policies and CP-SAT validators, bring the layout to a technologically feasible and economically best online solution.The traditional 3D case is NP-hard, hence the domination of very pointwise candidate constructive heuristics usable at predictable runtimes in operation improved by metaheuristics and learnable policies under tight time budgets, explicit stability and compatibility checking for physical feasibility changing the objective from proxy metrics-boxes or fill rate-to forecasted cost instead removes choice paradoxes and minimizes payments integration into a WMS with early rating, telemetry of actual dimensions and closed verification loop material consumption and trips which lowers carbon footprint while keeping throughput.The article will be helpful to WMS and fulfillment system architects, warehouse and transportation operations managers, combinatorial optimization researchers, e-commerce packing solution developers, and .NET stack integrators.