检索结果-内蒙古大学图书馆

Conference on computer Vision and Pattern Recognition (CVPR)

作者： Gengyu Zhang Hao Tang Yan Yan Department of Computer Science Illinois Institute of Technology USA Robotics Institute Carnegie Mellon University USA

ISBN: (数字)9798350353006

ISBN: (纸本)9798350353013

Route planning for navigation under partial observability plays a crucial role in modern robotics and autonomous driving. Existing route planning approaches can be categorized into two main classes: traditional autoregressive and diffusion-based methods. The former often fails due to its myopic nature, while the latter either assumes full observability or struggles to adapt to unfamiliar scenarios, due to strong couplings with behavior cloning from experts. To address these deficiencies, we propose a versatile diffusion-based approach for both 2D and 3D route planning under partial observability. Specifically, our value-guided diffusion policy first generates plans to predict actions across various timesteps, providing ample foresight to the planning. It then employs a differentiable planner with state estimations to derive a value function, directing the agent's exploration and goal-seeking behaviors without seeking experts while explicitly addressing partial observability. During inference, our policy is further enhanced by a best-plan-selection strategy, substantially boosting the planning success rate. Moreover, we propose projecting point clouds, derived from RGB-D inputs, onto 2D grid-based bird-eye- view maps via semantic segmentation, generalizing to 3D environments. This simple yet effective adaption enables zero-shot transfer from 2D-trained policy to 3D, cutting across the laborious training for 3D policy, and thus certifying our versatility. Experimental results demonstrate our superior performance, particularly in navigating situations beyond expert demonstrations, surpassing state-of-the-art autoregressive and diffusion-based baselines for both 2D and 3D scenarios.

关键词： Training Point cloud compression Three-dimensional displays Uncertainty Navigation Semantic segmentation Planning

来源：评论

学校读者我要写书评

暂无评论

Modality-Driven Design for Multi-Step Dexterous Manipulation: Insights from Neuroscience

arXiv

引用

arXiv 2024年

作者： Wake, Naoki Kanehira, Atsushi Saito, Daichi Takamatsu, Jun Sasabuchi, Kazuhiro Koike, Hideki Ikeuchi, Katsushi Applied Robotics Research Microsoft RedmondWA United States Department of Computer Science Institute of Science Tokyo Tokyo Japan

Multi-step dexterous manipulation is a fundamental skill in household scenarios, yet remains an underexplored area in robotics. This paper proposes a modular approach, where each step of the manipulation process is addressed with dedicated policies based on effective modality input, rather than relying on a single end-to-end model. To demonstrate this, a dexterous robotic hand performs a manipulation task involving picking up and rotating a box. Guided by insights from neuroscience, the task is decomposed into three sub-skills—reaching, grasping and lifting, and in-hand rotation—based on the dominant sensory modalities employed in the human brain. Each sub-skill is addressed using distinct methods from a practical perspective: a classical controller, a Vision-Language-Action model, and a reinforcement learning policy with force feedback, respectively. We tested the pipeline on a real robot to demonstrate the feasibility of our approach. The key contribution of this study lies in presenting a neuroscience-inspired, modality-driven methodology for multi-step dexterous manipulation. Copyright © 2024, The Authors. All rights reserved.

关键词： Visual languages

来源：评论

学校读者我要写书评

暂无评论

On the 3D trochoidal motion model of LiDAR sensors placed off-centered inside spherical mobile mapping systems

On the 3D trochoidal motion model of LiDAR sensors placed of...

引用

IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS)

作者： Fabian Arzberger Andreas Nüchter Computer Science XVII – Robotics Julius-Maximilians-Universität Würzburg Am Hubland Germany

ISBN: (数字)9798350377705

ISBN: (纸本)9798350377712

We study the motion model of a sensor rigidly mounted inside a ball. Due to the rigid placement inside the ball, the geometry of the sensor trajectory resembles a 3D curate trochoid. A new calibration method for spherical systems estimates the extrinsic parameters of the sensor with respect to the balls center of rotation. We deploy the calibration and motion model on our spherical mobile mapping platform to estimate the trajectory of a LiDAR sensor and compare it to trajectories of state-of-the-art LiDAR-Inertial odometry (LIO) methods. The motion model, which is solely based on IMU measurements, produces comparable results to the LIO methods, sometimes even outperforming them in positional accuracy. Although the LIO methods provide better rotational accuracy due to the utilization of LiDAR data, they struggle to reproduce the trochoidal nature of the trajectory and only provide pose estimations at the LiDAR frequency, whereas the motion model produces a more consistent trochoidal trajectory at the much higher IMU frequency. The results demonstrate the difficulty that current LIO methods have on spherical systems and indicate that our motion model is suitable for overcoming these issues.

关键词： Point cloud compression Geometry Laser radar Image color analysis Predictive models Three-dimensional printing Frequency estimation Data models Trajectory Calibration

来源：评论

学校读者我要写书评

暂无评论

Analysis of Causal Factor of Driver Fatigue in the Driving in Different Driving Conditions

Analysis of Causal Factor of Driver Fatigue in the Driving i...

引用

International Joint Conference on Neural Networks (IJCNN)

作者： Masataka Adachi Sou Nobukawa Keiichiro Inagaki Department of Robotics Science and Engineering Chubu University Kasugai Japan Department of Computer Science Chiba Institute of Technology Narashino Japan Department of Artificial Intelligence and Robotics Chubu University Kasugai Japan

ISBN: (数字)9798350359312

ISBN: (纸本)9798350359329

Addressing driving fatigue is crucial among the various factors in traffic accidents. The origins of driver fatigue encompass prolonged driving, sleep deprivation, and the induction of stress inherent in driving. The causes of such driver fatigue were categorized into physical factors, including sleep deprivation and prolonged driving, and perceptional-behavioral factors such as perception of driving conditions. Numerous studies have explored technologies to detect driver fatigue in terms of relevance to sleep states, eye movements, electroencephalography (EEG) activation levels, and their functional connectivity. It was also found differences in the occurrence of driver fatigue and related theta and alpha band activity based on driving conditions. However, the factors contributing to these differences remained unexplored. This study seeks to investigate the causal relationship between driver fatigue induced in different driving environments and the corresponding brain activity. We focus on perceptional load, particularly the perception of traffic objects within the driving condition, and physical loads, such as steering, acceleration, and braking, associated with variations in driving environments. In the experiment, we evaluate the relationship between such differences in physical and mental load and the occurrence of fatigue dependent on the driving condition with considering EEG activation patterns. In the results, we found that the amount of load related to driver perception and operation inherent in driving environments plays a crucial role in driver fatigue.

关键词： Sleep Human factors Fatigue Electroencephalography Biological neural networks Accidents

来源：评论

学校读者我要写书评

暂无评论

Resilient Consensus in Opinion Dynamics Under Adversarial Epidemics

引用

IFAC-PapersOnLine 2023年第2期56卷 1841-1846页

作者： Tatsuya Masada Yuan Wang Hideaki Ishii Department of Computer Science Tokyo Institute of Technology Yokohama Japan Department of Robotics Hunan University Changsha China

In this paper, we discuss the opinion dynamics with bounded confidence in multi-agent systems under an infection spreading environment. The dynamics of the infection spreading processes follows the so-called susceptible-infected-recovered (SIR) model. Here, the infection induces faulty behaviors in the agents whose opinions may deviate from their true opinions. Cooperating with infection suppression policies and the resilient algorithm based on the mean sub-sequence reduced (MSR) approach, resilient consensus can be attained by the regular agents within a safe region. In particular, we establish sufficient conditions for resilient consensus of opinion dynamics with large bounded confidence. A numerical example is provided to verify the effectiveness of our proposed approach.

关键词： Multi-agent systems Opinion dynamics Resilient consensus SIR epidemic model

来源：评论

学校读者我要写书评

暂无评论

Passively Adapting External Force Compensation System for Serial Manipulators 16th

Passively Adapting External Force Compensation System for S...

引用

16th International Federation of Theory of Machines and Mechanisms World Congress, IFToMM WC 2023

作者： Demian, Albert Klimchik, Alexandr Center for Technologies in Robotics and Mechatronics Components Innopolis University Innopolis420500 Russia School of Computer Science College of Science University of Lincoln Brayford Pool Lincolnshire LincolnLN6 7TS United Kingdom

ISBN: (纸本)9783031457692

This paper proposes a design approach for a spring-based compensator that passively adapts to external loadings. The proposed compensator is based on a combination of spring-lever mechanisms that are adjustable to the robot’s configuration. Springs in this arrangement are mounted on sliding pivots that connected to springs that allow passive adjustment of the value of the counter-torque. While previous research is either concerned with gravity compensation or with variable stiffness actuators, the proposed approach deals with the variation of external loadings. The proposed force compensator can be used in robotics applications where the robot experiences high external loads during operation. The model was tested in simulation for a planar 2-DoF manipulator and a planar redundant 3-DoF manipulator. The presented results show 58.5% torque reduction for 2-DoF and 55.2% reduction for 3-DoF. Robot redundancy can enhance counter-balancing since the compensation level is configuration-dependent. © 2023, The Author(s), under exclusive license to Springer Nature Switzerland AG.

关键词： Robots

来源：评论

学校读者我要写书评

暂无评论

Simulation of a Differential Robot Drive in Gazebo

Simulation of a Differential Robot Drive in Gazebo

引用

2023 International Conference "Quality Management, Transport and Information Security, Information Technologies", IT and QM and IS 2023

作者： Tseeva, Fatimat M. Shogenova, Mariyana M. Senov, Khamisha M. Nogerov, Ibragim A. Bozieva, Asiyat M. Institute of Computer Science Electronics and Robotics Federal State Budgetary Educational Institution of Higher Education Kabardino-Balkarian State University named after H.M. Berbekov Nalchik Russia

ISBN: (纸本)9798350340532

A differential drive is one of the most common drives in robotics that drives a robot by rotating two wheels at different speeds. This article discusses the features of modeling a robot differential drive in the Gazebo simulator. In particular, the key challenges that designers face when modeling differential drive, such as motion control and wheel inertia modeling, will be presented, as well as solutions that can be applied to achieve the desired results. The article will also describe how to use ROS (Robot Operating System) to simulate differential drive in Gazebo. ROS provides tools for controlling a robot in a simulation and provides a convenient way to exchange data between the components of a robot system. Finally, several examples will be presented showing how Gazebo can be used to simulate various types of differential drive robots, including wheeled robots and mobile robots with wide tracks. These examples demonstrate how proper simulation of a differential drive can be useful for developing and testing robot control algorithms in simulation, which can greatly speed up the development process and reduce the cost of experimental work. © 2023 IEEE.

关键词： Robot Operating System

来源：评论

学校读者我要写书评

暂无评论

Human-Robot Interaction Studies with Adults in Health and Wellbeing Contexts - Outcomes and Challenges 15th

Human-Robot Interaction Studies with Adults in Health and...

引用

15th International Conference on Social robotics, ICSR 2023

作者： Ghafurian, Moojan Dautenhahn, Kerstin Teka, Arsema Chandra, Shruti Rasouli, Samira Baliyan, Ishan Hutchinson, Rebecca Social and Intelligent Robotics Research Laboratory Faculty of Engineering University of Waterloo WaterlooON Canada David R. Cheriton School of Computer Science University of Waterloo WaterlooON Canada Davis Centre Library University of Waterloo WaterlooON Canada

ISBN: (纸本)9789819987146

Social robots have great potential to support individuals’ health and wellbeing. We did a follow up study based on previous work of a 2021 systematic review following PRISMA guidelines that identified 443 research articles evaluating social robots in health/wellbeing contexts with adult participants. In this paper we performed a new analysis that showed that while the vast majority of the articles reported positive outcomes related to the use of social robots in these contexts, only half of those articles supported the results by statistical tests and comparisons, highlighting a need for future studies with robust methodologies that can further study and inform the use of social robots for supporting adults in health/wellbeing. We discuss that different qualitative and quantitative methodologies are equally valuable, as long as the conclusions are based on the data collected. We also encourage publication of studies with well designed and executed methodologies that lead to neutral or negative outcomes. This is in line with other scientific research fields that emphasize the need to report on such results to avoid needless replication of studies that have already been done and could provide important lessons for the field, but never got published. © 2024, The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd.

关键词： Human robot interaction

来源：评论

学校读者我要写书评

暂无评论

Federated Learning in Competitive EV Charging Market

Federated Learning in Competitive EV Charging Market

引用

2023 IEEE PES Innovative Smart Grid Technologies Europe, ISGT EUROPE 2023

作者： Sun, Chenxi Huang, Chao Shou, Biying Huang, Jianwei Shenzhen Institute of Artificial Intelligence and Robotics for Society Shenzhen China Department of Computer Science University of California Davis United States Department of Management Science City University of Hong Kong Hong Kong Hong Kong Shenzhen China

ISBN: (纸本)9798350396782

Federated Learning (FL) has demonstrated a significant potential to improve the quality of service (QoS) of EV charging stations. While existing studies have primarily focused on developing FL algorithms, the effect of FL on the charging stations’ operation in terms of price competition has yet to be fully understood. This paper aims to fill this gap by modeling the strategic interactions between two charging stations and EV owners as a multi-stage game. Each station first decides its FL participation strategy and charging price, and then individual EV owners decide their charging strategies. The game analysis involves solving a non-concave problem and by decomposing it into a piece-wise concave program we manage to fully characterize the equilibrium. Based on real-world datasets, our numerical results reveal an interesting insight: even if FL improves QoS, it can lead to smaller profits for both stations. The key reason is that FL intensifies the price competition between charging stations by improving stations’ QoS to a similar level. We further show that the stations will participate in FL when their data distributions are mildly dissimilar. © 2023 IEEE.

关键词： charging stations competitive pricing Electric vehicles (EV) federated learning (FL) game theory

来源：评论

学校读者我要写书评

暂无评论

Securing the Internet of Things: Exploring Ensemble Learning for Attacks Classification

Securing the Internet of Things: Exploring Ensemble Learning...

引用

Industrial Engineering and Artificial Intelligence (IEAI), International Conference on

作者： Maurel Kouekam Fadoua Khennou Perception Robotics and Intelligent Machines Lab Computer Science departement Université de Moncton Moncton Canada

ISBN: (数字)9798350386363

ISBN: (纸本)9798350386370

Keeping the Internet of Things (IoT) safe from cyber threats is a big challenge, and Intrusion Detection Systems (IDS) are like security guards in this effort. While IoT facilitates the connection of diverse devices to provide intelligent services, it must guard against threats such as identity impersonation, service disruption, signal interference, and covert surveillance. As time progresses, attacks evolve, and new threats emerge on a daily basis. Consequently, the implementation of an Intrusion Detection System (IDS) becomes essential for the timely detection of potential threats on this platform. This study explores the potential effectiveness of machine learning algorithms in identifying attacks using the CICIoT2023 dataset. Our research aims to identify optimal classifiers for IoT attacks by leveraging ensemble learning methods to achieve the highest possible performance. Through experimentation, we have determined that LightGBM outperforms all other state-of-the-art machine learning models, achieving a 91% f1-score in the multiclassification of 8 attack categories.

关键词： Machine learning algorithms Surveillance Intrusion detection Interference Industrial engineering Feature extraction Internet of Things

来源：评论

学校读者我要写书评

暂无评论

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案：

请选择收藏分类：

通借通还

建议与咨询 留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案： 新增检索档案 确定 取消

请选择收藏分类： 新增自定义分类 确定 取消

通借通还

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

请选择保存的检索档案：

请选择收藏分类：