检索结果-南通市图书馆

Convergence analysis of an incremental approach to online inverse reinforcement learning

维普期刊数据库评论

在线全文

维普期刊数据库

学校读者我要写书评

暂无评论

Journal of Zhejiang University-Science C(Computers and Electronics) 2011年第1期12卷 17-24页

作者： Zhuo-jun JIN Hui QIAN Shen-yi CHEN Miao-liang ZHU School of Computer Science and Technology Zhejiang University Hangzhou 310027 China

Interest in inverse reinforcement learning (IRL) has recently increased,that is,interest in the problem of recovering the reward function underlying a Markov decision process (MDP) given the dynamics of the system and the behavior of an *** paper deals with an incremental approach to online ***,the convergence property of the incremental method for the IRL problem was investigated,and the bounds of both the mistake number during the learning process and regret were provided by using a detailed *** an online algorithm based on incremental error correcting was derived to deal with the IRL *** key idea is to add an increment to the current reward estimate each time an action mismatch *** leads to an estimate that approaches a target optimal *** proposed method was tested in a driving simulation experiment and found to be able to efficiently recover an adequate reward function.

关键词： Incremental approach Reward recovering Online learning Inverse reinforcement learning Markov decision process

A new approach of formation navigation derived from multi-robots cooperative online FastSLAM

在线全文

学校读者我要写书评

暂无评论

控制理论与应用（英文版） 2012年第4期10卷 451-457页

作者： Yunfei CAI Zhenmin TANG Chunxia ZHAO Institute of Pattern Recognition and Artificial Intelligence Nanjing University of Science and Technology

A new formation navigation approach derived from multi-robots cooperative online FastSLAM is proposed. In this approach, the leader and follower robots are defined. The posteriori estimation of the leader robot state is treated as a relative reference for all follower robots to correct their state priori estimations. The control volume of individual follower will be achieved from the results of the corrected estimation. All robots are observed as landmarks with known associations by the others and are considered in their landmarks updating. By the method, the errors of the robot posterior estimations are reduced and the formation is well kept. The simulation and physical experiment results show that the multi-robots relative localization accuracy is improved and the formation navigation control is more stable and efficient than normal leader-following strategy. The algorithm is easy in implementation.

关键词： Multi-robots Formation navigation FastSLAM

Using concurrent lines in central catadioptric camera calibration

在线全文

学校读者我要写书评

暂无评论

Journal of Zhejiang University-Science C(Computers and Electronics) 2011年第3期12卷 239-249页

作者： Lei ZHANG Xin DU Ji-lin LIU Department of Information Science and Electronic Engineering Zhejiang University Hangzhou 310027 China

Central catadioptric cameras have been extensively adopted in robotics and surveillance due to their extensive field of *** attain precise 3D information in these applications,it is important to calibrate the catadioptric cameras *** existing calibration techniques either require prior knowledge of the mirror types,or highly depend on a conic estimation procedure,which might be ruined if there are only small portions of the conic visible on calibration *** this paper,we design a novel planar pattern with concurrent lines as a calibration rig,which is more robust in conic estimation since the relationship among lines is taken into *** on the line properties,we propose a rough-to-fine approach suitable for the new planar pattern to calibrate central catadioptric *** method divides the nonlinear optimization calibration problem into several linear sub-problems that are much more robust against *** calibration method can estimate intrinsic parameters and the mirror parameter simultaneously and accurately,without a priori knowledge of the mirror *** performance is demonstrated by both simulation and a real hyperbolic catadioptric imaging system.

关键词： Catadioptric Calibration Concurrent lines Conic estimation

Design of an enhanced visual odometry by building and matching compressive panoramic landmarks online

在线全文

学校读者我要写书评

暂无评论

Frontiers of Information Technology & Electronic Engineering 2015年第2期16卷 152-165页

作者： Wei LU Instttute of lnformat:on and Communtcatlon Engineering Zhejtang Umvers:ty Hangzhou 310027 China Zhejiang Provincial Key Laboratory of Information Network Technology Hangzhou 310027 China

研究目的:高效精确定位是移动机器人智能导航的先决条件。传统视觉定位系统,如视觉里程计(VO)和同时定位与三维重建(SLAM)算法,存在两点不足:一是由累积定位误差引起的漂移问题,二是由光照变化和移动物体导致的错误运动估计结果。创新要... 详细信息

研究目的:高效精确定位是移动机器人智能导航的先决条件。传统视觉定位系统,如视觉里程计(VO)和同时定位与三维重建(SLAM)算法,存在两点不足:一是由累积定位误差引起的漂移问题,二是由光照变化和移动物体导致的错误运动估计结果。创新要点:通过引入全景相机到传统双目VO系统,提出一种增强型VO,高效利用全景相机360°视场角信息。(1)在线建立路口场景压缩全景路标库;(2)机器人以任意方向重新访问路标时,对定位结果进行全局校正;(3)当双目立体VO不能提供可靠定位信息时对航向角估计结果进行校正;(4)为高效利用信息量较多的全景图像,引入压缩感知概念并提出一种自适应压缩特征。研究方法:首先,在压缩亮度特征基础上,增加压缩SURF特征提高其描述能力,通过分析特征区分度,使压缩特征可以根据具体图像特点自适应调节,最终构建自适应压缩特征(ACF,图2),该特征计算速度快(表3)、描述能力强(图6、7,表1),有效提高全景图像信息利用效率。然后,使用ACF对全景路标图像进行描述,提出一种任意方向的路标图像匹配算法,若当前全景图像与路标图像匹配成功,则对当前定位结果进行全局位姿校正(图4),抑制大范围环境中定位路径漂移问题(图10、11)。最后,介绍基于图像片匹配的航向角鲁棒估计方法,当双目视觉里程计因特征跟踪质量差而导致运动估计结果不稳定时,对局部运动估计结果进行校正,提高运动估计的精度(图9)。重要结论:提出的增强型视觉里程计系统可以准实时提供可靠定位结果,极大抑制大范围挑战性环境中传统VO漂移问题和运动估计错误问题。实验结果显示,所提算法大幅度提高传统VO的准确性和鲁棒性。

关键词：视觉里程计全景路标路标匹配压缩感知自适应压缩特征

Unwrapping and stereo rectification for omnidirectional images

在线全文

学校读者我要写书评

暂无评论

Journal of Zhejiang University-Science A(Applied Physics & Engineering) 2009年第8期10卷 1125-1139页

作者： Jie LEI Xin DU Yun-fang ZHU Ji-lin LIU Institute of lnformation and Communication Engineering Zhejiang University Hangzhou 310027 China Zhejiang Provincial Key Laboratory of lnformation Network Technology Hangzhou 310027 China College of Computer and lnformation Engineering Zhejiang Gongshang University Hangzhou 310027 China

Omnidirectional imaging sensors have been used in more and more applications when a very large field of view is *** this paper,we investigate the unwrapping,epipolar geometry and stereo rectification issues for omnidirectional vision when the particular mirror model and the camera parameters are unknown in ***,the omnidirectional camera is calibrated under the Taylor model,and the parameters related to this model are *** order to make the classical computer vision algorithms of conventional perspective cameras applicable,the ring omnidirectional image is unwrapped into two kinds of panoramas:cylinder and *** the epipolar geometry of arbitrary camera configuration is analyzed and the essential matrix is deduced with its properties being indicated for ring *** that,a simple stereo rectification method based on the essential matrix and the conformal mapping is *** and real data experimental results illustrate that our methods are effective for the omnidirectional camera under the constraint of a single view point.

关键词：立体图像整改计算机视觉算法摄像机参数基本矩阵图像传感器摄像机校准立体视觉

New Layered SOA-Based Architecture for Multi-robots Cooperative Online SLAM

在线全文

学校读者我要写书评

暂无评论

Chinese Journal of Electronics 2014年第1期23卷 25-30页

作者： CAI Yunfei TANG Zhenmin ZHAO Chunxia Institute of Pattern Recognition and Artificial Intelligence Nanjing University of Science and Technology

A new layered multi-robots architecture for cooperative online simultaneous localization and mapping based on Service-oriented architecture(SOA) is proposed. This architecture is constructed by the concepts of service-oriented and loosely coupled center *** methods make robots' underlying function realization be transparent in cooperation and avoid the impact affected by robot heterogeneous characteristics, which will be beneficial to the system construction, expansion, restructuring and maintenance. All data packages follow the SOA unified data format, which can avoid the impact affected by sensor heterogeneous characteristics. The point to point serving model reduces the path of data transmission which can improve the timeliness of the system. All of these features can satisfy the requirements of multi-robots cooperative online SLAM. The simulation results show the feasibility, applicability and practicability of this architecture.

关键词： Service-oriented architecture(SOA) Multi-robot cooperative online SLAM(MRCO-SLAM) Loosely coupled

A new feature fusion method at decision level and its application

在线全文

同方期刊数据库

学校读者我要写书评

暂无评论

Optoelectronics Letters 2010年第2期6卷 129-132页

作者：韩光赵春霞张浩峰袁夏 College of Computer Science and Technology Nanjing University of Science and Technology

To solve the problem that using only one feature is poor for the terrestrial environment classification,a new feature fusion method at decision level is proposed in this *** eigenvector is obtained by using a Gaussian mixture model(GMM) firstly,and the probabilities of the eigenvector belonging to a certain class of the texture and color are *** the probabilities are multiplied by different weights according to the contribution,and summed to get the maximal likelihood probability to achieve feature *** results demonstrate that the method in this paper is better than other current methods and the classification performance is superior to a single feature obviously.

关键词：融合方法决策层应用高斯混合模型特征向量概率计算环境分类 GMM

在线全文

学校读者我要写书评

暂无评论

交互式多区域模型

电子学报 2012年第6期40卷 1235-1239页

作者：曲彦文张二华杨静宇江西师范大学计算机信息工程学院江西南昌330000 南京理工大学计算机科学与技术学院江苏南京210094

一种被称为交互式多区域模型(IMRM)的非线性滤波算法被提出,用于对状态和连续系统参数进行联合估计.IMRM将连续的系统参数空间视为由若干子区域所构成的集合,并将每个子区域分别分配给一个子模型.IM-RM使用一组子滤波器并行滤波.在每一... 详细信息

一种被称为交互式多区域模型(IMRM)的非线性滤波算法被提出,用于对状态和连续系统参数进行联合估计.IMRM将连续的系统参数空间视为由若干子区域所构成的集合,并将每个子区域分别分配给一个子模型.IM-RM使用一组子滤波器并行滤波.在每一时刻,IMRM利用交互操作计算各子模型的混合初始化环境,之后各子滤波器在假设系统参数跳变到特定子区域的前提下,对状态和系统参数进行估计.为了有效地应用IMRM,提出了一种基于无迹变换的交互式多区域模型(UT-IMRM)算法.UT-IMRM对每个子模型使用无迹卡尔曼滤波器(UKF)进行滤波.在目标跟踪实验中对UT-IMRM性能进行测试.实验结果显示当系统参数不属于IMM模型集合时,UT-IMRM能够比IMM获得更好的估计性能.

关键词：交互式多模型参数估计无迹卡尔曼滤波器

基于Cross-EKF定位的多机器人协作围捕策略研究

在线全文

学校读者我要写书评

暂无评论

控制与决策 2010年第9期25卷 1313-1317,1323页

作者：蔡云飞唐振民张浩峰南京理工大学计算机科学与技术学院南京210094

针对目前多机器人协作围捕过程中收敛速度慢、稳定性差、定位精度低的问题,提出一种新的围捕策略.设计出Cross-EKF定位算法,对目标位置的后验估计协方差进行交叉计算,以取得最小协方差区域.以区域边缘点到均值中心最大距离为半径,构建... 详细信息

针对目前多机器人协作围捕过程中收敛速度慢、稳定性差、定位精度低的问题,提出一种新的围捕策略.设计出Cross-EKF定位算法,对目标位置的后验估计协方差进行交叉计算,以取得最小协方差区域.以区域边缘点到均值中心最大距离为半径,构建收敛圆,将对动态点的收敛扩展为对动态面的收敛.实验结果表明,系统能快速平稳地收敛该圆,从而实现对目标的精确围捕,该方法具有较高的实用价值.

关键词：多机器人 Cross-EKF定位协作围捕