检索结果-内蒙古大学图书馆

Data-driven Output-feedback Predictive Control: Unknown Plant's Order and Measurement Noise

学校读者我要写书评

暂无评论

arXiv 2022年

作者： Jo, Nam H. Shim, Hyungbo Department of Electrical Engineering Soongsil University Seoul Korea Republic of ASRI Department of Electrical and Computer Engineering Seoul National University Seoul Korea Republic of

The aim of this paper is to propose a new data-driven control scheme for multi-input-multi-output linear time-invariant systems whose system model are completely unknown. Using a non-minimal input-output realization, the proposed method can be applied to the case where the system order is unknown, provided that its upper bound is known. A workaround against measurement noise is proposed and it is shown through simulation study that the proposed method is superior to the conventional methods when dealing with input/output data corrupted by measurement noise. © 2022, CC BY-NC-SA.

关键词： Spurious signal noise

A Reference Shaping Filter Design Method for Discrete-Time Sliding Mode Control with Decoupled Disturbance Compensator to Inhibit Overshoot

学校读者我要写书评

暂无评论

A Reference Shaping Filter Design Method for Discrete-Time S...

International Conference on Control, Automation and Systems ( ICCAS)

作者： Dae-Young Yang Tae-Ho Oh Young-Seok Kim Tae-Hoon Kim Sang-Hoon Lee Dong-Il Dan Cho Dept. of Electrical and Computer Engineering ASRI/ISRC Seoul National University Seoul Korea RS Automation Co. Ltd. Gyeonggi-do Korea

The discrete-time sliding mode control (DSMC) is a highly utilized method in industrial systems, such as manufacturing systems, robot manipulators, and automation systems. The DSMC with decoupled disturbance compensator (DDC) method is one of the robust control methods with high tracking performance and fast responses. Based on error dynamics, a performance-shaping methodology for DSMC with DDC has also been developed to reduce tracking position errors in industrial servo systems. However, the input reference to the output transfer function of the DSMC has a transfer function zero, causing an unwanted overshoot and degrading the performance of the controller. Thus, this overshoot characteristic can be problematic in some systems that have stringent no-overshoot specifications. In this paper, a reference shaping filter design method is proposed that inhibits any overshoot in the response of the DSMC with DDC in a double-integrator system. The presented controller is applied to a linear motor system to illustrate the performance.

关键词：

TOWARDS REALISTIC INCREMENTAL SCENARIO IN CLASS INCREMENTAL SEMANTIC SEGMENTATION

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Kwak, Jihwan Cha, Sungmin Moon, Taesup Department of Electrical and Computer Engineering Seoul National University Korea Republic of Computer Science Department Courant Institute of Mathematical Sciences New York University United States ASRI INMC IPAI AIIS Seoul National University Korea Republic of

This paper addresses the unrealistic aspect of the overlapped scenario, a commonly adopted incremental learning scenario in Class Incremental Semantic Segmentation (CISS). We highlight that the overlapped scenario allows the same image to reappear in future tasks with different pixel labels, creating unwanted advantage or disadvantage to widely used techniques in CISS, such as pseudo-labeling and data replay from the exemplar memory. Our experiments show that methods trained and evaluated under the overlapped scenario can produce biased results, potentially affecting algorithm adoption in practical applications. To mitigate this, we propose a practical scenario called partitioned, where the dataset is first divided into distinct subsets representing each class, and then these subsets are assigned to corresponding tasks. This efficiently addresses the data reappearance artifact while meeting other requirements of CISS scenario, such as capturing the background shifts. Additionally, we identify and resolve the code implementation issues related to replaying data from the exemplar memory, previously overlooked in other works. Lastly, we introduce a simple yet competitive memory-based baseline, MiB-AugM, that handles background shifts in the exemplar memory. This baseline achieves state-of-the-art results across multiple tasks involving learning many new classes. Codes are available at https: //***/jihwankwak/CISS_partitioned. Copyright © 2024, The Authors. All rights reserved.

关键词： Semantic Segmentation

DefectFill: Realistic Defect Generation with Inpainting Diffusion Model for Visual Inspection

学校读者我要写书评

暂无评论

arXiv 2025年

作者： Song, Jaewoo Park, Daemin Baek, Kanghyun Lee, Sangyub Choi, Jooyoung Kim, Eunji Yoon, Sungroh Department of Electrical and Computer Engineering Seoul National University Korea Republic of Global Technology Research Samsung Electronics Korea Republic of IPAI Korea Republic of AIIS ASRI INMC ISRC Seoul National University Korea Republic of

Developing effective visual inspection models remains challenging due to the scarcity of defect data. While image generation models have been used to synthesize defect images, producing highly realistic defects remains difficult. We propose DefectFill, a novel method for realistic defect generation that requires only a few reference defect images. It leverages a fine-tuned inpainting diffusion model, optimized with our custom loss functions incorporating defect, object, and attention terms. It enables precise capture of detailed, localized defect features and their seamless integration into defect-free objects. Additionally, our Low-Fidelity Selection method further enhances the defect sample quality. Experiments show that DefectFill generates high-quality defect images, enabling visual inspection models to achieve state-of-the-art performance on the MVTec AD dataset. Copyright © 2025, The Authors. All rights reserved.

关键词： Defects

Wasserstein Distributionally Robust Control and State Estimation for Partially Observable Linear Systems

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Jang, Minhyuk Hakobyan, Astghik Yang, Insoon Department of Mechanical Engineering Seoul National University Seoul Korea Republic of Center for Scientific Innovation and Education National Polytechnic University of Armenia Yerevan Armenia Department of Electrical and Computer Engineering and ASRI Seoul National University Seoul Korea Republic of

This paper presents a novel Wasserstein distributionally robust control and state estimation algorithm for partially observable linear stochastic systems, where the probability distributions of disturbances and measurement noises are unknown. Our method consists of the control and state estimation phases to handle distributional ambiguities of system disturbances and measurement noises, respectively. Leveraging tools from modern distributionally robust optimization, we consider an approximation of the control problem with an arbitrary nominal distribution and derive its closed-form optimal solution. We show that the separation principle holds, thereby allowing the state estimator to be designed separately. A novel distributionally robust Kalman filter is then proposed as an optimal solution to the state estimation problem with Gaussian nominal distributions. Our key contribution is the combination of distributionally robust control and state estimation into a unified algorithm. This is achieved by formulating a tractable semidefinite programming problem that iteratively determines the worst-case covariance matrices of all uncertainties, leading to a scalable and efficient algorithm. Our method is also shown to enjoy a guaranteed cost property as well as a probabilistic out-of-sample performance guarantee. The results of our numerical experiments demonstrate the performance and computational efficiency of the proposed method. © 2024, CC BY-NC-ND.

关键词： State estimation

Meta-Explore: Exploratory Hierarchical Vision-and-Language Navigation Using Scene Object Spectrum Grounding

学校读者我要写书评

暂无评论

arXiv 2023年

作者： Hwang, Minyoung Jeong, Jaeyeon Kim, Minsoo Oh, Yoonseon Oh, Songhwai Electrical and Computer Engineering and ASRI Seoul National University Korea Republic of Department of Electronic Engineering Hanyang University Korea Republic of Interdisciplinary Major in Artificial Intelligence Seoul National University Korea Republic of

The main challenge in vision-and-language navigation (VLN) is how to understand natural-language instructions in an unseen environment. The main limitation of conventional VLN algorithms is that if an action is mistaken, the agent fails to follow the instructions or explores unnecessary regions, leading the agent to an irrecoverable path. To tackle this problem, we propose Meta-Explore, a hierarchical navigation method deploying an exploitation policy to correct misled recent actions. We show that an exploitation policy, which moves the agent toward a well-chosen local goal among unvisited but observable states, outperforms a method which moves the agent to a previously visited state. We also highlight the demand for imagining regretful explorations with semantically meaningful clues. The key to our approach is understanding the object placements around the agent in spectral-domain. Specifically, we present a novel visual representation, called scene object spectrum (SOS), which performs category-wise 2D Fourier transform of detected objects. Combining exploitation policy and SOS features, the agent can correct its path by choosing a promising local goal. We evaluate our method in three VLN benchmarks: R2R, SOON, and REVERIE. Meta-Explore outperforms other baselines and shows significant generalization performance. In addition, local goal search using the proposed spectral-domain SOS features significantly improves the success rate by 17.1%andSPLby 20.6%againstthe state-of-the-art method of the SOON benchmark. Project page: https://***/projects/Meta-Explore/*** © 2023, CC BY-NC-SA.

关键词： Navigation

Unseen Road Type Detection in Road Networks for Intelligent Transportation Systems

学校读者我要写书评

暂无评论

Unseen Road Type Detection in Road Networks for Intelligent ...

International Conference on Intelligent Transportation

作者： Daeho Um Yuneil Yeo Ji Won Yoon Jin Young Choi Department of Electrical and Computer Engineering Automation and Systems Research Institute (ASRI) Seoul National University Seoul South Korea Department of Civil and Environmental Engineering University of California Berkeley USA Department of Artificial Intelligence Chung-Ang University Seoul South Korea

ISBN: (数字)9798331505929

ISBN: (纸本)9798331505936

Existing methods for road type (highway, trunk road, etc.) classification assume that every road in real environments belongs to a road type seen during training. However, there are unseen road types in real-world scenarios. Thus, unreliable classification of an unseen-type road into a seen road type can cause critical safety issues in road-related applications. In this paper, we introduce a new framework to detect unseen road types. To this end, we adopt an out-of-distribution (OOD) detection approach studied in the deep learning field. However, conventional graph-based node-level OOD detection methods cannot be directly applied to the unseen road type detection problem since roads are represented by edges in road networks. To resolve this problem, we establish a new formulation of edge-level OOD detection and propose a novel energy propagation scheme on a line graph transformed from a road network to obtain OOD scores. Experimental results on real-world road networks demonstrate the effectiveness of our method, achieving state-of-the-art performance in unseen road type detection.

关键词： Training Deep learning Roads Image edge detection Energy resolution Safety Intelligent transportation systems

RE-WEIGHTING BASED GROUP FAIRNESS REGULARIZATION VIA CLASSWISE ROBUST OPTIMIZATION

学校读者我要写书评

暂无评论

arXiv 2023年

作者： Jung, Sangwon Park, Taeeon Chun, Sanghyuk Moon, Taesup Department of Electrical and Computer Engineering Seoul National University Korea Republic of NAVER AI Lab United States ASRI/INMC IPAI AIIS Seoul National University Korea Republic of

Many existing group fairness-aware training methods aim to achieve the group fairness by either re-weighting underrepresented groups based on certain rules or using weakly approximated surrogates for the fairness metrics in the objective as regularization terms. Although each of the learning schemes has its own strength in terms of applicability or performance, respectively, it is difficult for any method in the either category to be considered as a gold standard since their successful performances are typically limited to specific cases. To that end, we propose a principled method, dubbed as FairDRO, which unifies the two learning schemes by incorporating a well-justified group fairness metric into the training objective using a classwise distributionally robust optimization (DRO) framework. We then develop an iterative optimization algorithm that minimizes the resulting objective by automatically producing the correct re-weights for each group. Our experiments show that FairDRO is scalable and easily adaptable to diverse applications, and consistently achieves the state-of-the-art performance on several benchmark datasets in terms of the accuracy-fairness trade-off, compared to recent strong baselines. Copyright © 2023, The Authors. All rights reserved.

关键词： Benchmarking

On Concentration Bounds for Bayesian Identification of Linear Non-Gaussian Systems

学校读者我要写书评

暂无评论

On Concentration Bounds for Bayesian Identification of Linea...

IEEE Conference on Decision and Control

作者： Yeoneung Kim Gihun Kim Insoon Yang Department of Applied Artificial Intelligence Seoul National University of Science and Technology Seoul South Korea Department of Electrical and Computer Engineering ASRI Seoul National University Seoul South Korea

We adopt a Bayesian perspective to identify the unknown parameters of linear stochastic systems with possibly non-Gaussian disturbance distributions. The key idea of our algorithm is to alternately execute $L$ randomly selected linear state-feedback controllers and keep track of a maximum a posteriori estimator. The proposed algorithm asymptotically achieves the concentration of posterior distributions around the true system parameters. We also derive probabilistic bounds for the concentration based on the classical results regarding the asymptotic properties of posterior distributions. An empirical demonstration is provided as well.

关键词：