• <tr id="yyy80"></tr>
  • <sup id="yyy80"></sup>
  • <tfoot id="yyy80"><noscript id="yyy80"></noscript></tfoot>
  • 99热精品在线国产_美女午夜性视频免费_国产精品国产高清国产av_av欧美777_自拍偷自拍亚洲精品老妇_亚洲熟女精品中文字幕_www日本黄色视频网_国产精品野战在线观看 ?

    Service Robot Localization Based on Global Vision and Stereo Vision

    2012-02-07 07:46:48YUQingxiao于清曉YANWeixin閆維新FUZhuangZHAOYanzheng趙言正

    YU Qing-xiao(于清曉),YAN Wei-xin(閆維新),F(xiàn)U Zhuang(付 莊),ZHAO Yan-zheng(趙言正)*

    1 State Key Laboratory of Mechanical System and Vibration,Shanghai Jiaotong University,Shanghai 200240,China

    2 State Key Laboratory of Robotics and System,Harbin Institute of Technology,Harbin 150001,China

    Introduction

    Mobile robot localization and object pose estimation in a working environment have been central research activities in mobile robotics.Solution of mobile robot localization needs to address two main problems[1]:the robotmusthave a representation of the environment;the robot must have a representation of its belief regarding its pose in this environment.Sensors are the basis of addressing both problems.Many of the sensors available to mobile robots are introduced in Ref.[1],giving their basic principles and performance limitations.Based on this introduction,ultrasonic sensors[2,3],goniometers[4],laser range finders[5,6],and charge coupled device(CCD)cameras[7,8]are sensors which are commonly applied in mobile robot localization projects to gather high precision information on the robot's pose estimation.

    Most mobile robot localization approaches can be widely classified into three major types:metric,topological,and hybrid.Metric approaches[9,10]are useful when it is necessary for the robot to know its position accurately in terms of metric coordinates.And,the state of the robot can also be represented in a more qualitative manner,by using a topological map[11,12].Butthese methods confrontmany difficulties when the environments are changed.A vision-based localization and mapping algorithm which used scale-invariant image feature(STIF)was applied for mobile robot localization and map building in Ref.[13].Miro et al.[14]used the binocular vision system to generate the disparity map by matching feature and realized the self-localization and self-navigation.Grzyb et al.[15]adopted stereo images and biologically inspired algorithms to accurately estimate the pose,size,and shape of the target object.Greggio et al.[16]applied stereo cameras and a real-time tracking algorithm to realize pattern recognition and evaluate the 3D position of a generic spatial object.Chinellato et al.[17]proposed a visual analysis which was based on a computational model of distance and orientation estimation inspired by human visual mechanisms to extract the features and pose of the object.

    Considering the positioning cost and computational efficiency,a hierarchical localization method based on vision is proposed to provide the absolute coordinates in this paper.The proposed algorithm can estimate the real-time coordinates well,and obtain up to±2 cm positioning accuracy to make sure that the robot can successfully grab the object.

    1 Hierarchical Positioning Method

    1.1 Global vision-based coarse localization

    The estimation of the mobile robot's pose is a fundamental problem,which can be roughly divided into two classes[13]:methods for keeping track of the robot's pose and methods for global pose estimation.Most ofthe present research has concentrated on the first class,which assumes that the initial pose of the robot is known[18].In this study,both methods are used to position the mobile robot in the initial stage.The global vision-based localization method is used to calculate the initial position oftherobotand periodically update itscurrent coordinates,and the dead-reckoning method is also used to localize the robot as a supplementary method.In the following part,the global vision-based localization method is explained in detail.

    One monocular camera mounted on the ceiling is used to acquire and transmit the global image to the robot through the wireless network.The image is firstly processed to remove the image distortion.Then,color segmentation method is used to position the robot in the image and distinguish the color mark on the head in the Hue-Saturation-Value(HSV)color space.After distinguishing and extracting the color mark,the coordinates of the color mark can be obtained to calculate the coordinates of the robot,according to the global vision model shown in Fig.1.

    In Fig.1,(0,0)denotes the origin of the camera coordinate system;HCdenotes the height of the camera relative to the floor;HRdenotes the height of the robot;(XRC,YRC)is the center coordinate of the robot;(XHC,YHC)is the center coordinate of the color mark.

    According to the geometric relationship,the following equation can be obtained:

    Then,the following results can be calculated as:

    Fig.1 The global vision camera model

    In Eq.(2),we know that(XRC,YRC)is dependent on the values of HC,HR,and(XHC,YHC).But the values of HCand HRare invariant in this study,and(XRC,YRC)is only dependent on(XHC,YHC).

    Four coordinate systems:global coordinate system(GCS),robot coordinate system(RCS),sensor coordinate system (SCS),and virtualglobalcoordinate system(VGCS)[19]are all under consideration.So,the appropriate transformation between VGCS and SCS should be acquired to calculate the absolute coordinates of the robot.According to the geometric relationship in Fig. 1, the homogeneous transformation matrix T between OXYZ and O′X′Y′Z′can be obtained as:

    Then,the coordinates of the robot in VGCS can be calculated as:

    In the study,VGCS is coincident with GCS in addition to different scale.So,the absolute coordinates of the robot can be achieved as:

    where Kheightand Kwidthare the relative scales to the VGCS;m and n are shift distances of VGCS relative to SCS.

    In this way,the robot can know its initial position wherever it is placed and periodically update its current coordinates.Furthermore,the real-time positioning of the robot can be realized in indoor environment with the help of the deadreckoning method.

    1.2 Stereo vision-based precise localization

    When the service robot moves into the area where the object is placed under the guidance of the global vison,the binocular vision-based localization method is adopted to provide high positioning accuracy to successfully grab the object for the customer.In this study,color segmentation and shape-based matching[20]are separately applied to detect the object,and the extracted results are fused to obtain the accurate object.The color segmentation method has been discussed.Below,we describe the shape-based matching method.

    The advantage ofshape-based matching isitsgreat robustness and flexibility.Instead of using the gray values,features along contours are extracted and used for the model generation and matching.The method is invariant to changes in illumination and variations of object gray values.The method also allows the object to be rotated and scaled.The process of shape-based matching is divided into two distinct phases:the training phase and the recognition phase.In the first phase,the template model should be created and trained.In the second phase,the model is used to find and localize the object in the image.The basic concept of shape-based image matching is shown in Fig.2.

    Fig.2 The basic concept of shape-based image matching

    Using the mentioned mehods,the regions of interest(ROI)are seperately obtained. Then,the following information fusion method is adopted to find the expected region of the object.Suppose that image regions A are obtained using color segmentation method and image regions B are also obtained using shape-based matching method.The center points of image regions A and B are all extracted and compared one by one.If the difference between the center coordinates of one A image region and the center coordinates of one B image region is less than the default constant ε,the image area belongs to the expected regions of interest.Otherwise,the image area is the error detected region.Then,the expected regions of interest can be obtained as:

    where(xi,yi)is the center coordinates of the sub-region of the image regions A;(xj,yj)is the center coordinates of the j subregion of the image regions B.

    According to Eq.(6),the expected regions can be obtained to find and match the object.After obtaining the plate region,the normalized cross-correlation(NCC)matching method[21]is adopted to obtain the depth image.Then,the 3D coordinates of the color mark,which is on the object,can be calculated using the binocular vision system.To localize the mobile robot,the homogeneous transformation matrix between binocular coordinate system and RCS should be firstly obtained.Figure 3 shows the relationship between binocular coordinate system and RCS.

    Fig.3 The diagram of binocular coordinate system and RCS

    In Fig.3,α and θ are the rotation degrees of the head;O′X′Y″Z denotesRCS,and O″X″Y″Z″ denotesbinocular coordinate system.When α and θ are equal to zero,O″X″Y″Z″shifts negative 6 cm in the X-axis direction relative to O′X′Y′Z′.According to these information, the homogeneous transformation matrix T between O′X′Y′Z′and O″X″Y″Z″can be got as:

    Then,the coordinates of the color mark in RCS can be calculated as :

    where[x′ y′ z′]Tis the location of the color mark in RCS;[x″ y″ z″]Tis the location of the color mark in binocular coordinate system.

    In addition to obtain the 3D coordinates of the color mark,the angle of the object relative to the robot should be also calculated by extracting and filtering the edge lines of the object.Detailed process is as follows.Firstly,the region of the object is processed using the expansion algorithm and canny operator.Then,the extracted edge lines are filtered using the threshold filter to obtain the available edge lines which can denote the angle of the object.Then,the selected edge lines are filtered using the mean filter,which can be used to calculate the angle of the object.Finally,the angle of the object relative to the robot,φobject,can be reckoned as follows.

    where φ denotes the detected angle of the object in the image and θ is the rotation angle of the neck joint.

    After determining the position of the color mark,the robot can adjust the velocity and heading angle to move towards the object under the guidance of the location-based visual servo control system.Periodically localize the position of the color mark using the method above and adjust the pose of the robot until the robot successfully realizes its precise positioning relative to the color mark.The location-based visual servo control system is shown in Fig.4.

    Fig.4 The chart of the location-based visual servo control system

    2 Experiments and Result Analysis

    2.1 Experiment setup

    The experimental platform is a restaurant robot as shown in Fig.5 which is equipped with one binocular camera.The model of the binular camera is bumblebee2 which is made by Point Grey Research.There is a global camera mounted on the ceiling which is used to obtain the whole environmental image.The global images with 768×576 pixels are sent to the mobile robot through the wirless ethernet.The binocular camera mounted on the robot head can send the stereo images with 512×384 piexls to the robot through 1 394 bus.Before the experiment,the binocular camera has been calibrated in advance.

    The complete movement of the service robot can be divided into two parts:coarse localization and precise localization.In the following parts,we explain the experimental results of the complete movement from the two aspects.

    Fig.5 The photo of the restaurant service robot

    2.2 Global vision-based coarse localization

    In this part,the global vision localization method can be explained through the following example in this study.The mobile robot receives the image of the restaurant from the global camera through the wirless ethernet.To acquire the available value from the image,we should firstly rectify the raw image.Figure 6 is the rectified global image obtained from the camera.To localize the robot in the image,we should find the ROI and detect the color mark on the robot head using the color segmentation method.The result of the image processing is shown in Fig.7.

    After the location of the robot is detected in the image,the location of the robot relative to the pinhole camera coordinate system should be calculated.We have known the size of the picture in advance.And the height of the camera relative to the floor and the height of the robot are invariant and also known in advance.Because the position and the parameters of the global camera are invariant,the scales of the pixel relative to the one centimeter in the X-axis and Y-axis directions can be acquired.These parameters which are used to calculate the location of the robot,are all shown in Table 1.

    Table 1 The used parameters in this experiment

    In Fig.7,the origin of the picture coordinate system is in the top left-hand corner of the picture.And the coordinates of the color mark in the picture coordinate system have been calculated.According to Fig.1,we obtain the transformation coordinates,(X'HC,Y'HC),in the pinhole coordinate system.Then,the result can be calculated as(-68.50,- 311.33).After(X'HC,Y'HC)is obtained,the robot can be positioned in the image as(-39.14,-177.90).

    Next,the real coordinates of the robot in GCS should be reckoned.The origin of GCS is in the top right-hand of the carpet array shown in Fig.7.And,the coordinates of the top right-hand of the carpet array is(216,237)in the pinhole coordinate system.So,the coordinates of the robot in GCS is calculated as

    Many experiment results show that the accuracy of global vision positioning is within ±7 cm.In this study,the coordinates of the robot is updated every 100 ms to make sure that the robot can obtain its real-time coordinates.

    2.3 Stereo vision-based precise localization

    When the robot moves into the area where the object is placed under the guidance of the global vison,the stereo visionbased localization method is adopted toprovide the high positioning accuracy.The acquired images from binocular camera are firstly rectified using image correction algorithm to prepare the available images for the next process.The rectified image from the stereo vision during the robot movement is shown in Fig.8.

    Fig.8 The rectified stereo vision image

    In this study,the image taken by the right camera is used as the reference image where the ROI is extracted to realize the object feature matching and obtain the disparity map of the object.The object region is detected and extracted using color segmentation and shape-based matching methods.The procedure of color segmentation is as follows.Firstly,HSV color spaces could be obtained through the image color space conversion.Then,the appropriate thresholds are set in the Hue and Saturation space components.Finally,the plate region is extracted from the reference image.And the plate region can be also found in the image using the shape-based matching method.Each center coordinates of the extracted regions of interest can be firstly extracted,then,the expected region can be obtained according to Eq.(6).The plate region in Fig.9 is the expected region which is obtained according to the methods above.

    Fig.9 The expected region of interest

    Next,we should find the characteristic sign in the expected region and match image using NCC algorithm.In the process,the consistency test between the left and right images is also used to find the accuracy matching.The obtained depth image is shown in Fig.10.

    Fig.10 The obtained depth image using NCC

    From Fig.10,we can know that the depth value of the characteristic sign is 40.07.According to the binocular vision system,the coordinates of the characteristic mark can be calculated as[1.4 0.2 118.9]T.Otherwise,using the encoders,we can know that α =45°and θ=0°.Then,using Eq.(7),the homogeneous transformation matrix T,between RCS and binocular vision coordinate system is denoted as follows:

    Finally,according to Eq.(8),the result can be reckoned as[x′ y′ z′]=[- 4.6 88.5 83.9].Otherwise,the angle of the object relative to the robot should be calculated to determine the position of the color mark.In experiment,firstly,the edge lines of the object are extracted using canny operator.Then,the extracted edge lines are filtered using the threshold filter to find the available lines.Finally,the selected lines are filtered using the mean filter to calculate the angle of the object.The obtained lines using the method above are shown in Fig.11.From Fig.11,we know that the angle of the object is-3.865°.

    Fig.11 The extracted edge lines of the object

    After determining the position of the color mark,the robot can adjust the velocity and heading angle to move towards the object under the guidance of the location-based visual servo control system.And in the process of the movement,the robot should constantly repeat the positioning process above and adjust its pose until the precise positioning is finally realized.

    We performed the whole movement at the velocity of 0.15 m/s for sixty times in our library,and the initial position of the robot was random and unknown each time.And,the service robot could always realize self-localization with the positioning accuracy up to±2 cm and successfully grab the object for the customers within 3 min.The statistical data are shown in Table 2.

    Table 2 The application and positioning accuracy of the hierarchical localization method

    3 Conclusions and Future Work

    In this paper,a method of mobile robot localization which is based on visual feedback is proposed and applied on the restaurant service robot.The whole positioning process can be divided into two stages:coarse positioning and precise positioning.Hierarchical positioningmethod is adopted to provide different positioning accuracies in different stages.In the coarse positioning stage,the robot can periodically obtain its current coordinates using the global vision-based localization method.In the precise positioning stage,the robot can obtain higher localization precision using the binocular vision-based localization method.Finally,the robot can successfully grasp the plate on the table.Many experiments verify that the proposed algorithm has good location effect and obtains up to±2 cm positioning accuracy.

    Currently,research on the improvement in the positioning accuracy and speed of the algorithm is under way.And our future work will also focus on improving the anti-interference ability to the more complex environment.

    [1]Siegwart R,Nourbakhsh I.Introduction to Autonomous Mobile Robots[M].Cambridge:The MIT Press,2004:121-156.

    [2]Choi B S,Lee J J.Mobile Robot Localization Scheme Based on RFID andSonarFusion System [C].IEEE International Symposium on Industrial Electronics,Seoul,Korea,2009:1035-1040.

    [3]Choi B S,Lee J J.Localization of a Mobile Robot Based on an Ultrasonic Sensor Using Dynamic Obstacles[J].Artificial Life and Robotics,2008,12(1/2):280-283.

    [4]Bonnifait P,Garcia G.Design and Experimental Validation of an Odometric and Goniometric Localization System for Outdoor Robot Vehicles[J].IEEE Transactionson Roboticsand Automation,1998,14(4):541-548.

    [5]Balaguer B,Carpin S,Balakirsky S. Towards Quantitative Comparisons of Robot Algorithms:Experiences with SLAM in Simulation and Real World Systems[C].IEEE/RSJ International Conference on Intelligent Robots and Systems,California,USA,2007:1-7.

    [6]Lin H H,Tsai C C.Laser Pose Estimation and Tracking Using Fuzzy Extended Information Filtering for an Autonomous Mobile Robot[J].Journal of Intelligent and Robotic Systems,2008,53(2):119-143.

    [7]Munguia R,Grau A.Monocular SLAM for Visual Odometry[C]. IEEE InternationalSymposium on IntelligentSignal Processing,Madrid,Spain,2007:1-6.

    [8]Lin H Y,Lin J H,Wang M L.A Visual Positioning System for Vehicle or Mobile Robot Navigation[C].Proceedings of the 8th InternationalIEEE Conference on Intelligent Trasportation Systems,Vienna,Austria,2005:73-78.

    [9]Blanco J L,Gonzalez J,F(xiàn)ernandez-Madrigal J A.Consistent Observation Grouping for Generating Metric-Topological Maps that Improves Robot Localization[C].Proceeding of IEEE International Conference on Robotics and Automation,Orlando,USA,2006:818-823.

    [10]Guo Y,Xu X H.Color Landmark Design for Mobile Robot Localization[C].IMACS Multiconference on Computational Engineering in Systems Applications,Beijing,China,2006:1868-1874.

    [11]Kwon T B,Yang J H,Song J B,et al.Efficiency Improvement in Monte Carlo Localization through Topological Information[C].Proceeding of 2006 IEEE/RSJ International Conference on Intelligent Robots and Systems,Beijing,China,2006:424-429.

    [12]Tapus A,Siegwart R.A Cognitive Modeling of Space Using Fingerprints ofPlaces for Mobile Robot Navigation [C].Proceedings of the IEEE International Conference on Robotics and Automation,Orlando,USA,2006:1188-1193.

    [13]Borenstein J,Everett B,F(xiàn)eng L.Navigation Mobile Robot:System and Techniques[M].Wellesley,USA:A.K.Peters Ltd.,1996:103-150.

    [14]Miro J V,Zhou W Z,Dissanayake G.Towards Vision Based Navigation in Large Indoor Environments[C]. IEEE International Conference on Intelligent Robots and Systems,Beijing,China,2006:2096-2102.

    [15]Grzyb B,Chinellato E,Morales A,et al.A 3D Grasping System Based on MultimodalVisualand TactileProcessing [J].Industrial Robot:An International Journal,2009,36(4):365-369.

    [16]Greggio N,Bernardino A,Laschi C,et al.Real-Time 3D Stereo Tracking and Localizing of Spherical Objects with the iCub Robotic Platform [J].Journal of Intelligent and Robotic Systems,2011,63(3/4):417-446.

    [17]Chinellato E,Grzyb B J,del Pobil A P.Brain Mechanisms for Robotic ObjectPose Estimation[C]. InternationalJoint Conference on Neural Networks,Hong Kong,China,2008:3268-3275.

    [18]Lang H X,Wang Y,de Silva C W.Mobile Robot Localization and Object Pose Estimation Using Optical Encoder,Vision and Laser Sensors[C].Proceedings of the IEEE International Conference on Automation and Logistics,Qingdao,China,2008:617-622.

    [19]Yu Q X,Yuan C,F(xiàn)u Z,et al.Research of the Localization of Restaurant Service Robot[J].International Journal of Advanced Robotic Systems,2010,7(3):227-238.

    [20]El Munim H A,F(xiàn)arag A A. Shape Representation and Registration Using Vector Distance Functions[C].Proceedings of theIEEE Conference on ComputerVision and Pattern Recognition,Minneapolis,USA,2007:1-8.

    [21]Sun Z X,Wu Q.TS201 Based Fast Algorithm of Normalized Cross-Correlation[J].Modern Electronics Technique,2010,33(10):125-127.(in Chinese)

    好男人在线观看高清免费视频| 国产精品人妻久久久久久| 国产成+人综合+亚洲专区| 99久久久亚洲精品蜜臀av| www.色视频.com| 听说在线观看完整版免费高清| 18禁裸乳无遮挡免费网站照片| 国产亚洲欧美98| a级一级毛片免费在线观看| 久久精品91蜜桃| 久久国产乱子免费精品| 91在线精品国自产拍蜜月| 97超级碰碰碰精品色视频在线观看| 99久久精品一区二区三区| 999久久久精品免费观看国产| 精品一区二区免费观看| 性色avwww在线观看| 亚洲国产精品久久男人天堂| 欧美不卡视频在线免费观看| 日韩欧美一区二区三区在线观看| 欧美黑人巨大hd| 91麻豆av在线| 一个人看的www免费观看视频| 亚洲精品一区av在线观看| 欧美一区二区国产精品久久精品| 亚洲最大成人手机在线| 精品国内亚洲2022精品成人| 又粗又爽又猛毛片免费看| 免费高清视频大片| 国产美女午夜福利| 在线观看舔阴道视频| 国产三级中文精品| 欧美日韩中文字幕国产精品一区二区三区| 69av精品久久久久久| 国产亚洲欧美98| 91九色精品人成在线观看| 亚洲av成人不卡在线观看播放网| a级毛片a级免费在线| 18禁黄网站禁片午夜丰满| 久久久久久久亚洲中文字幕 | 欧美xxxx黑人xx丫x性爽| 中文字幕免费在线视频6| 99国产精品一区二区三区| 亚洲av成人不卡在线观看播放网| 乱人视频在线观看| 亚洲精品456在线播放app | 亚洲精品一区av在线观看| a级毛片a级免费在线| 久久久色成人| 日日摸夜夜添夜夜添小说| 每晚都被弄得嗷嗷叫到高潮| or卡值多少钱| 如何舔出高潮| 国语自产精品视频在线第100页| 亚洲乱码一区二区免费版| 久久婷婷人人爽人人干人人爱| 搡女人真爽免费视频火全软件 | 欧美日韩乱码在线| 国产伦人伦偷精品视频| 97热精品久久久久久| 在线免费观看不下载黄p国产 | 欧美国产日韩亚洲一区| 成人一区二区视频在线观看| 久久精品国产亚洲av涩爱 | 成人国产一区最新在线观看| 午夜a级毛片| 91久久精品国产一区二区成人| 特大巨黑吊av在线直播| 在线观看66精品国产| 一区福利在线观看| 免费观看的影片在线观看| 国产又黄又爽又无遮挡在线| 三级男女做爰猛烈吃奶摸视频| 久久精品久久久久久噜噜老黄 | 老司机午夜福利在线观看视频| 特级一级黄色大片| www.色视频.com| 十八禁人妻一区二区| 午夜久久久久精精品| 午夜两性在线视频| 亚洲精品粉嫩美女一区| 成人鲁丝片一二三区免费| 久久精品国产亚洲av涩爱 | 久久久久久久亚洲中文字幕 | 最新在线观看一区二区三区| 免费无遮挡裸体视频| 男女做爰动态图高潮gif福利片| 男女视频在线观看网站免费| 国产成人a区在线观看| 精品一区二区三区人妻视频| 国产一区二区激情短视频| 热99re8久久精品国产| 午夜免费男女啪啪视频观看 | 日韩中文字幕欧美一区二区| 久久久久久国产a免费观看| 91狼人影院| 亚洲精品在线观看二区| 欧美在线一区亚洲| 国产av一区在线观看免费| 国产视频一区二区在线看| 亚洲av免费在线观看| 国产欧美日韩一区二区精品| 赤兔流量卡办理| av在线观看视频网站免费| 少妇裸体淫交视频免费看高清| 欧美黄色淫秽网站| 亚洲专区国产一区二区| 国产成人福利小说| 亚洲欧美激情综合另类| 免费av不卡在线播放| 亚洲av免费在线观看| 熟女人妻精品中文字幕| 亚洲最大成人中文| 亚洲激情在线av| 毛片女人毛片| 免费人成在线观看视频色| 久久欧美精品欧美久久欧美| 久久亚洲真实| 国产激情偷乱视频一区二区| 亚洲精品成人久久久久久| 老鸭窝网址在线观看| 日韩中文字幕欧美一区二区| 久久人人爽人人爽人人片va | 日本黄色视频三级网站网址| 极品教师在线免费播放| 免费黄网站久久成人精品 | 亚洲人与动物交配视频| 伊人久久精品亚洲午夜| 97超视频在线观看视频| 久久99热6这里只有精品| 国产毛片a区久久久久| 久久精品夜夜夜夜夜久久蜜豆| 色精品久久人妻99蜜桃| 国产探花在线观看一区二区| 日韩成人在线观看一区二区三区| 免费观看的影片在线观看| 免费一级毛片在线播放高清视频| av国产免费在线观看| 中国美女看黄片| 国产色婷婷99| 久久欧美精品欧美久久欧美| 最近最新免费中文字幕在线| 淫妇啪啪啪对白视频| 日韩人妻高清精品专区| 成人午夜高清在线视频| 精品国内亚洲2022精品成人| 国产不卡一卡二| 日本一二三区视频观看| 国产伦在线观看视频一区| 久久精品国产亚洲av香蕉五月| 日本一二三区视频观看| 日韩国内少妇激情av| 久久久久久久久大av| av女优亚洲男人天堂| 国产精品不卡视频一区二区 | 国内精品久久久久精免费| 日韩精品青青久久久久久| 最近中文字幕高清免费大全6 | 精品国内亚洲2022精品成人| 在线观看一区二区三区| 最近在线观看免费完整版| 亚洲午夜理论影院| 亚洲激情在线av| 国产成人aa在线观看| 欧美+日韩+精品| 黄色女人牲交| 99热这里只有是精品在线观看 | 午夜精品久久久久久毛片777| 亚洲中文日韩欧美视频| 精品一区二区三区人妻视频| 国产精品女同一区二区软件 | 丰满人妻熟妇乱又伦精品不卡| 国内精品美女久久久久久| 最近视频中文字幕2019在线8| 午夜福利成人在线免费观看| 欧美日韩瑟瑟在线播放| 麻豆成人av在线观看| 精品久久久久久成人av| 久久国产精品影院| 亚洲av成人av| 久久午夜亚洲精品久久| 99精品在免费线老司机午夜| 一区二区三区高清视频在线| 国产高清视频在线观看网站| 成人性生交大片免费视频hd| 啦啦啦观看免费观看视频高清| 久久精品91蜜桃| 午夜福利免费观看在线| 99久久精品国产亚洲精品| 国产精品国产高清国产av| 亚洲国产精品sss在线观看| 超碰av人人做人人爽久久| 如何舔出高潮| 亚洲人成网站在线播放欧美日韩| 美女黄网站色视频| 亚洲欧美日韩高清专用| 午夜视频国产福利| 夜夜看夜夜爽夜夜摸| 我的女老师完整版在线观看| 嫩草影院入口| 高清在线国产一区| 国产探花极品一区二区| 村上凉子中文字幕在线| 制服丝袜大香蕉在线| 久久草成人影院| 一级黄色大片毛片| 高清日韩中文字幕在线| 午夜精品在线福利| av专区在线播放| 日本黄色视频三级网站网址| 女生性感内裤真人,穿戴方法视频| 欧美日韩中文字幕国产精品一区二区三区| 欧美zozozo另类| 直男gayav资源| 一个人看视频在线观看www免费| 1000部很黄的大片| 亚洲18禁久久av| 国产中年淑女户外野战色| 国产av麻豆久久久久久久| 欧美精品啪啪一区二区三区| 国内精品美女久久久久久| 九九热线精品视视频播放| 久99久视频精品免费| 亚洲欧美精品综合久久99| 激情在线观看视频在线高清| 自拍偷自拍亚洲精品老妇| 噜噜噜噜噜久久久久久91| 久久精品国产99精品国产亚洲性色| 老熟妇乱子伦视频在线观看| avwww免费| 青草久久国产| 日日夜夜操网爽| 2021天堂中文幕一二区在线观| 国产黄色小视频在线观看| 欧美高清性xxxxhd video| 小说图片视频综合网站| 精品久久国产蜜桃| 色综合欧美亚洲国产小说| 国产老妇女一区| 精品午夜福利在线看| 嫁个100分男人电影在线观看| 国产欧美日韩精品一区二区| 精品一区二区三区人妻视频| 色噜噜av男人的天堂激情| 搡老妇女老女人老熟妇| 亚洲不卡免费看| 激情在线观看视频在线高清| 久久精品久久久久久噜噜老黄 | 天堂av国产一区二区熟女人妻| 欧美日韩黄片免| 精品午夜福利在线看| 好男人在线观看高清免费视频| 美女被艹到高潮喷水动态| 神马国产精品三级电影在线观看| 亚洲狠狠婷婷综合久久图片| 亚洲一区高清亚洲精品| 伊人久久精品亚洲午夜| eeuss影院久久| 欧美成人a在线观看| 99热这里只有是精品50| 精品一区二区三区人妻视频| 成人特级av手机在线观看| 成人鲁丝片一二三区免费| 国产免费一级a男人的天堂| 91久久精品国产一区二区成人| 日本一本二区三区精品| 精品一区二区三区视频在线观看免费| 久久精品国产自在天天线| 精品免费久久久久久久清纯| 成年免费大片在线观看| 国产在线男女| 成人国产一区最新在线观看| 深爱激情五月婷婷| 如何舔出高潮| 午夜两性在线视频| 久久精品影院6| 国产视频内射| 人妻制服诱惑在线中文字幕| 久久精品国产自在天天线| 中国美女看黄片| 99精品在免费线老司机午夜| 欧美日韩综合久久久久久 | 精品福利观看| 91麻豆精品激情在线观看国产| 精品人妻熟女av久视频| 变态另类成人亚洲欧美熟女| 毛片一级片免费看久久久久 | 欧美最新免费一区二区三区 | 亚洲av中文字字幕乱码综合| 十八禁网站免费在线| 尤物成人国产欧美一区二区三区| 美女大奶头视频| 在线免费观看不下载黄p国产 | www.999成人在线观看| 国内精品久久久久久久电影| 精品一区二区三区av网在线观看| 男人的好看免费观看在线视频| 欧美日韩福利视频一区二区| 色5月婷婷丁香| 亚洲美女黄片视频| 级片在线观看| 91在线观看av| 久9热在线精品视频| 国产探花极品一区二区| 久久亚洲精品不卡| 久久午夜亚洲精品久久| 真人做人爱边吃奶动态| 91字幕亚洲| 国产伦人伦偷精品视频| 丰满人妻熟妇乱又伦精品不卡| 999久久久精品免费观看国产| a级毛片免费高清观看在线播放| 午夜激情欧美在线| 亚洲,欧美,日韩| 五月伊人婷婷丁香| 久久午夜亚洲精品久久| 久久久久免费精品人妻一区二区| 五月伊人婷婷丁香| 一本久久中文字幕| 成人精品一区二区免费| 久久中文看片网| 欧洲精品卡2卡3卡4卡5卡区| 国产主播在线观看一区二区| 亚洲美女搞黄在线观看 | 人妻制服诱惑在线中文字幕| 国产精品爽爽va在线观看网站| 久久久精品大字幕| 国产成年人精品一区二区| 国产av一区在线观看免费| 亚洲片人在线观看| 国产久久久一区二区三区| 亚洲精品日韩av片在线观看| 亚洲精品456在线播放app | 国产成年人精品一区二区| 精品人妻一区二区三区麻豆 | 成人高潮视频无遮挡免费网站| 亚洲狠狠婷婷综合久久图片| 中文字幕av在线有码专区| 国产精品,欧美在线| 日韩欧美在线二视频| 3wmmmm亚洲av在线观看| 亚洲久久久久久中文字幕| 日韩中字成人| 午夜福利成人在线免费观看| 国内揄拍国产精品人妻在线| 别揉我奶头 嗯啊视频| av欧美777| 成年女人毛片免费观看观看9| 婷婷精品国产亚洲av在线| 亚洲人与动物交配视频| 国产三级中文精品| 在线看三级毛片| 老熟妇仑乱视频hdxx| 亚洲av免费高清在线观看| 熟女电影av网| 真人一进一出gif抽搐免费| 全区人妻精品视频| 久久性视频一级片| 亚洲一区二区三区色噜噜| 男人的好看免费观看在线视频| 熟女电影av网| 毛片女人毛片| 欧美一区二区精品小视频在线| 国产精品嫩草影院av在线观看 | 精品人妻1区二区| 午夜激情福利司机影院| 国产精品久久久久久亚洲av鲁大| 国产日本99.免费观看| 午夜免费成人在线视频| 欧美丝袜亚洲另类 | 听说在线观看完整版免费高清| 免费av观看视频| 国产精品爽爽va在线观看网站| 成人精品一区二区免费| 国产精品精品国产色婷婷| 在线播放无遮挡| 国产精品自产拍在线观看55亚洲| 亚洲,欧美,日韩| 性色av乱码一区二区三区2| 欧美+日韩+精品| 我要搜黄色片| 欧美不卡视频在线免费观看| 一区二区三区免费毛片| 亚洲电影在线观看av| 成年人黄色毛片网站| 欧美三级亚洲精品| 成人av在线播放网站| 在线观看av片永久免费下载| 亚洲欧美激情综合另类| 亚洲欧美日韩无卡精品| 久久久国产成人精品二区| 亚洲国产精品sss在线观看| 亚洲第一欧美日韩一区二区三区| 18禁黄网站禁片免费观看直播| 最新在线观看一区二区三区| 欧美在线黄色| av黄色大香蕉| 黄片小视频在线播放| av在线老鸭窝| 床上黄色一级片| 脱女人内裤的视频| 90打野战视频偷拍视频| 特级一级黄色大片| 亚洲欧美日韩卡通动漫| 又紧又爽又黄一区二区| 亚洲成av人片免费观看| 悠悠久久av| 国产精品美女特级片免费视频播放器| 日韩欧美国产一区二区入口| 国产单亲对白刺激| 久久99热这里只有精品18| 老师上课跳d突然被开到最大视频 久久午夜综合久久蜜桃 | 桃色一区二区三区在线观看| 国产一区二区在线观看日韩| 精品人妻一区二区三区麻豆 | 999久久久精品免费观看国产| 麻豆国产av国片精品| 悠悠久久av| 午夜激情欧美在线| 午夜福利18| 99久国产av精品| 日本 av在线| 精品无人区乱码1区二区| 我的老师免费观看完整版| 欧美又色又爽又黄视频| 亚洲激情在线av| 欧美日韩福利视频一区二区| 国产白丝娇喘喷水9色精品| 免费人成视频x8x8入口观看| 久久香蕉精品热| 国产乱人视频| 18禁在线播放成人免费| 乱码一卡2卡4卡精品| 九九久久精品国产亚洲av麻豆| 国产亚洲av嫩草精品影院| 18禁黄网站禁片午夜丰满| 精品国产亚洲在线| 亚洲欧美日韩卡通动漫| 欧美乱色亚洲激情| 九色成人免费人妻av| 尤物成人国产欧美一区二区三区| 亚洲成人中文字幕在线播放| 久久久国产成人精品二区| 日本免费一区二区三区高清不卡| 18美女黄网站色大片免费观看| 欧美日本视频| 五月玫瑰六月丁香| 欧美高清性xxxxhd video| av天堂在线播放| 99在线视频只有这里精品首页| 特级一级黄色大片| 变态另类丝袜制服| 国产精品一区二区免费欧美| 91麻豆精品激情在线观看国产| 又爽又黄a免费视频| 久久99热这里只有精品18| 日本三级黄在线观看| 久久精品久久久久久噜噜老黄 | 午夜a级毛片| 两人在一起打扑克的视频| 91久久精品电影网| 啪啪无遮挡十八禁网站| 色5月婷婷丁香| 亚洲精品粉嫩美女一区| 九色国产91popny在线| 久久精品久久久久久噜噜老黄 | 啦啦啦观看免费观看视频高清| 久久精品国产自在天天线| 在现免费观看毛片| 一级黄色大片毛片| 91久久精品国产一区二区成人| 久久久久久久久久成人| 乱人视频在线观看| 国产激情偷乱视频一区二区| 亚洲自偷自拍三级| 亚洲天堂国产精品一区在线| 少妇被粗大猛烈的视频| 夜夜夜夜夜久久久久| 嫩草影视91久久| 欧美日韩福利视频一区二区| 1000部很黄的大片| 免费观看的影片在线观看| 麻豆成人av在线观看| 亚洲五月天丁香| 国产精品一区二区三区四区免费观看 | 国产精品亚洲美女久久久| 夜夜夜夜夜久久久久| 动漫黄色视频在线观看| 女人十人毛片免费观看3o分钟| 亚洲久久久久久中文字幕| 精品久久久久久久久亚洲 | 亚洲成av人片免费观看| 国产视频一区二区在线看| 欧美成人一区二区免费高清观看| 伦理电影大哥的女人| 欧美zozozo另类| 久久久久国产精品人妻aⅴ院| 日本a在线网址| 99国产精品一区二区三区| 男女之事视频高清在线观看| 亚洲精品乱码久久久v下载方式| 成人毛片a级毛片在线播放| 一本久久中文字幕| 中文亚洲av片在线观看爽| 国产成人aa在线观看| 国产野战对白在线观看| 国产大屁股一区二区在线视频| a级毛片免费高清观看在线播放| 婷婷亚洲欧美| 简卡轻食公司| 男人舔女人下体高潮全视频| 成人一区二区视频在线观看| 两个人视频免费观看高清| 999久久久精品免费观看国产| 夜夜夜夜夜久久久久| 免费av毛片视频| 天堂av国产一区二区熟女人妻| 欧美激情在线99| 国产一区二区三区视频了| 性欧美人与动物交配| 黄片小视频在线播放| 中文资源天堂在线| 日韩欧美精品免费久久 | 精品99又大又爽又粗少妇毛片 | bbb黄色大片| 少妇丰满av| 免费人成在线观看视频色| 亚洲av第一区精品v没综合| 成年女人看的毛片在线观看| 国产亚洲欧美98| 欧美午夜高清在线| 国产亚洲精品av在线| 成人无遮挡网站| 夜夜看夜夜爽夜夜摸| 给我免费播放毛片高清在线观看| 国产三级中文精品| av黄色大香蕉| 国产一区二区在线av高清观看| 中文字幕av在线有码专区| 99国产综合亚洲精品| 在线观看66精品国产| 美女 人体艺术 gogo| www.色视频.com| 国产精品人妻久久久久久| 久久久久国内视频| 亚洲美女黄片视频| 精品一区二区三区av网在线观看| 亚洲av第一区精品v没综合| 不卡一级毛片| 91麻豆av在线| 国产一区二区激情短视频| 亚洲色图av天堂| 无人区码免费观看不卡| 国产欧美日韩一区二区三| 18+在线观看网站| 国产成人影院久久av| 欧美日本亚洲视频在线播放| 十八禁国产超污无遮挡网站| 国产黄色小视频在线观看| 人妻夜夜爽99麻豆av| 老司机福利观看| 国产一区二区亚洲精品在线观看| 黄色日韩在线| 久久久久性生活片| 亚洲,欧美精品.| 特级一级黄色大片| 国产欧美日韩一区二区精品| 午夜a级毛片| 在线国产一区二区在线| 99精品在免费线老司机午夜| 欧美在线一区亚洲| 日韩欧美国产一区二区入口| 国产又黄又爽又无遮挡在线| 午夜亚洲福利在线播放| 国产高清视频在线播放一区| 日本三级黄在线观看| 精品乱码久久久久久99久播| 精品国内亚洲2022精品成人| 嫩草影院精品99| 天天一区二区日本电影三级| a在线观看视频网站| 老熟妇仑乱视频hdxx| 99热6这里只有精品| 日韩中字成人| 国产又黄又爽又无遮挡在线| 变态另类丝袜制服| 欧美乱色亚洲激情| 麻豆国产av国片精品| 国产主播在线观看一区二区| 91九色精品人成在线观看| 亚洲最大成人手机在线| 午夜a级毛片| 中文资源天堂在线| 国产精品嫩草影院av在线观看 | 国产欧美日韩精品亚洲av| 少妇高潮的动态图| 老司机午夜十八禁免费视频| 国产91精品成人一区二区三区| 亚洲va日本ⅴa欧美va伊人久久| 亚洲人成电影免费在线| 久久婷婷人人爽人人干人人爱| 91狼人影院| 简卡轻食公司| 免费av毛片视频| 99国产精品一区二区三区| bbb黄色大片| 久久中文看片网| 黄色一级大片看看| 欧美色欧美亚洲另类二区| 深夜精品福利| 亚洲av二区三区四区| 国产精品亚洲av一区麻豆| 亚洲精品粉嫩美女一区| 九九在线视频观看精品| 成年女人毛片免费观看观看9|