靈巧機(jī)器人智能可視化操作界面

1、Proceedings of the 2000 IEEEIn ternational Conference on Robotics & Automation San Francisco, CA April 2000器人智能可視化操作界面t Vision-only Operator Interface for obotsMechanical designPartA HeadArmLegHandfreedom266Humanoid robotThe goal （本方案的目標(biāo)） Develop methods and algorithms enabling the natural and unenc

2、umbered generation of commands for anthropomorphic robots while minimizing specific operator training and dedicated equipment.:1、開發(fā)仿人機(jī)器人算法：在操作者接受盡量少的特 別訓(xùn)練和專門儀器的情況下，使仿人機(jī)器人命令的生 成正常而順暢。The project aims at allowing the use ofsophisticated, humalike robots in criticalenvironments, such as the Internation

3、al SpaceStation and the proposed Human Settlement onMars, where time, space and mass constraintsprevent the use of more traditional telepresencesystems, such as joysticks and force feedbackdevices.:2、致力于使復(fù)雜精密的仿人機(jī)器人在如：國(guó)際空間站 或已提出的人類登陸火星惡劣環(huán)境下可以工作，時(shí)間、 地點(diǎn)等大量約束都迫使我們不能再使用如操縱桿和力反 饋裝置等傳統(tǒng)的遙控系統(tǒng)。This work is mo

4、tivated by the parallel development, at NASA Johnson Space Center, of the anthropomorphic robotic system Robonaut, pursuing the design of a robot the size of a suited astronaut. This robot will be teleoperated by a non-intrusive telepresence interface, taking advantage of the natural physical corres

5、pondence between the robot and the human operator.這項(xiàng)工作由NASA JOHNSON空間中心設(shè)計(jì)的跟宇航員大小相 同的仿人機(jī)器人系統(tǒng)ROBONAUT的發(fā)展來推動(dòng)。這個(gè)機(jī)器人 將由無干擾遠(yuǎn)程監(jiān)控界面進(jìn)行遙操作，利用機(jī)器人和操作人員 之間的物理通信。teleoperator interfaces遙操作界面 1 The input system used here is a visual tracker of the operator arm motions輸入系統(tǒng)：對(duì)操作者手臂運(yùn)動(dòng)的視覺跟蹤器(visual tracker) 2 Graphical

6、 displays have been a constant feature of teleoperation output systems輸出系統(tǒng)：圖像顯示(Graphical displays)是遙操作輸出系統(tǒng) 中常見的形式。 3、Interface intelligence is usually embedded in the manipulator control algorithm with specific robustness and stability features 智能系統(tǒng)：界面的智能性(Interface intelligence)通常包含在 控制器中關(guān)于穩(wěn)定性的控制算

7、法內(nèi)-visual track視覺跟蹤In the past, visual tracking has been addressed with two main approaches過去視覺跟蹤主要采用兩種方式： marker-based systems 采用M ARKER的系統(tǒng) marker-less systems 不采用M ARKER的系統(tǒng)Marker-based systems usereflective markers on the body ofthe subject, infrared lighting andmulti-ple cameras.采用MARKER的系統(tǒng)利用紅外光和

8、多臺(tái)相機(jī)，將具有 反射作用的MARKER放在物體表面。此方法可以以非常高的精度計(jì)算M ARKER的三維空間 位置。The marker-less approach is basedon modeling the appearance ofbody parts in the images generatedby one or more cameras.不采用M ARKER的方法是基于由一架或多架相機(jī)生成 的圖像中身體各部分表面的模型。Graphical displays圖像顯示 Computer generated graphical simulations are used to give t

9、he operator status information and to compensatefor transmission time delay.采用計(jì)算機(jī)生成的圖像仿真將位置信息傳遞給操作者，并補(bǔ)償 傳輸時(shí)間延遲。 Intelligent interaction with the operator is addressed in this paper, where an interactive path planner for the teleoperation of dexterous manipulators is proposed文中提出了遙操作交互的途徑，使智能系統(tǒng)可以與操作者

10、進(jìn)行 A program monitors task execution and gives the operator feedback about task status程序監(jiān)測(cè)任務(wù)的執(zhí)行，將任務(wù)執(zhí)行的情況反饋給操作者。Interface intelligence界面的智能性 In terface intellige nee is usually embedded in the man ipulator control algorithm with specific robustness and stability features界面的智能性通常包含在控制器中具有精確和穩(wěn)定特點(diǎn)的控制算法中

11、。 For example, identification of kinematic singularities is critical for the correct teleoperation of a dexterous arm, since singularity location is not self-evident. In the past, singularity detection has been addressed primarily from the point of view of avoiding any unexpected and dangerous motio

12、n of the robot. In particular, approaches such as 15 solve the problem by forcing the robot to pass at a distanee from the nominal trajectory, whereas solutions such as 17 keep the robot velocity within acceptable limits. In a teleoperation system these changes can confuse the operator when they hap

13、pen without warn ings.例如，鑒別運(yùn)動(dòng)學(xué)奇異點(diǎn)對(duì)于正確操作靈巧臂是十分重要，但奇異點(diǎn)的位置 并不直觀。過去，奇異點(diǎn)的探測(cè)和定位，主要通過觀察，以避免機(jī)器人產(chǎn)生 任何的意外危險(xiǎn)的動(dòng)作。在一個(gè)遙操作系統(tǒng)內(nèi)這些變化在沒有預(yù)警的情況下 會(huì)干猊樣祚者。the proposed vision-basedtelepresence interface可視化遠(yuǎn)程監(jiān)控界面 The features of the Droposed vision-based elepresence interface are demonstrated in an off-lineimplementation ba

14、sed on iug the graphical simulation of a Robotics Research Corp. (RRC) 該特點(diǎn)被基于機(jī)器人研究 中心(RRC)的圖像仿真 禺線運(yùn)行崩論證lmaige Acquisitionk-Motion Estimation數(shù)據(jù)和步驟流程:Command Advisor如圖1所示Figure 1: Conceptual scheme of th interface.:逅1Command AdvisorImageAcquisiti-on:“WAI2.： r： Command GencraiioikFigure 1: Conceptual s

15、cheme of the interface.Motion Mim報(bào)io抵:操作者移動(dòng)手臂一覺追蹤提取關(guān)鍵點(diǎn)的笛卡兒坐將這些坐標(biāo)轉(zhuǎn)換為關(guān)節(jié)角分析關(guān)節(jié)角是否存在自身碰撞，計(jì)算與運(yùn)動(dòng)學(xué)奇異點(diǎn) 的距離一以圖像方式疊加在真實(shí)場(chǎng)景視頻圖像上，顯示給操作者本界面的整體結(jié)構(gòu)及其主要構(gòu)成2視覺跟蹤系統(tǒng) The Visual Tracker像顯示 The Graphical Display自身碰撞檢測(cè) SelfCollision Detection奇異空間的可視化 Visualization of Singularity SpaceThe Visual Tracker視覺跟蹤系統(tǒng):對(duì)操作者運(yùn)動(dòng)起視覺跟蹤作用的硬

16、件、軟件系統(tǒng):前端設(shè)備 動(dòng)作獲取設(shè)備 離線標(biāo)定軟件 3維重構(gòu)軟件 標(biāo)注軟件front-endacquisition hardware off-line calibration software 3D reconstruct!ng software labeling software前端設(shè)備 The front-end for measuring the operators arm motion is a non-invasive system based on a set of 4 videocameras and small markers placed on the main joints

17、 of the oper,ator5s body.由四個(gè)攝像頭和一些位于操作者身體主要關(guān)節(jié)的小MARKER組成 的無干擾系統(tǒng) (non-invasive system)。 The current implementation simplifies the problem of detecting the markers in each camera, using small light bulbs as markers and requiring a low level of illumination in the acquisition area 現(xiàn)行系統(tǒng)通過用小燈泡作為MARKER并在捕捉

18、區(qū)域內(nèi)采用極低的 照明將每個(gè)相機(jī)探測(cè)M ARKER的問題簡(jiǎn)化。Acquisition動(dòng)作獲取Figure 2: Block Diagram of the Acquisition System.We use 4 monochrome NTSC in terlacedcameras with 6mm lenses located on a semicircle around the center of the calibrated volume at a distance of about 5 m from the center and about 3 m above the floor leve

19、l. The automatic gain in each camera was disabled and the shutter time was set to 1/500 sec as a trade off between image contrast and motion blur.用4架鏡頭焦距為6毫米的單色NTSC相機(jī)，圍 繞標(biāo)定區(qū)域中心5米遠(yuǎn)、3米高圍成半圓。每架 相機(jī)不能自動(dòng)獲取圖像，快門時(shí)間為1/500秒 以達(dá)到畫面對(duì)比和運(yùn)動(dòng)模糊之間的平衡。Camera calibration相機(jī)的標(biāo)定internal parameters相機(jī)的相互位置mutual camera posit

20、ion世界參考坐標(biāo)系之間的嚴(yán)格轉(zhuǎn)換rigid transformation between cameras and world reference frameThe 3D reconstruct!on software3維重構(gòu)軟件The 3D reconstrucion software triangulatesbetween different views to determine the 3 dimensional position of each markers在不同視角之間做三角測(cè)量以測(cè)定每個(gè)M ARKER的3維位置。The visual tracking algorithm take

21、s the set of reconstructed 3D marker coordinates at each frame and assigns to each marker the correct body part The program requires t o manually label the 3D points in one of the frames in the sequenee.視覺追蹤算法采用重構(gòu)每禎MARKER的3維坐標(biāo)，然后將每個(gè) MARKER賦予正確的身體部位。程序需要在序列中的一禎對(duì)這 些3維的點(diǎn)進(jìn)行人工標(biāo)識(shí)。labeling software標(biāo)注軟件-c沁欝

22、沁、A,XVY,?J%,. 【 1 - * / r * I s: M i f fWi.Vir . . Figure 3: The Visualization/Editing Tool；1 XV滋I舄|諮 1 I The visualization tool shown in Figure 3 was implemented in Matlab t o provide a graphical representation of the computed data (for example representing the rec on structed body with a stick fig

23、ure), for inspection and editing.=1圖3所示為：可視化工具在 MATLAB下執(zhí)行，為計(jì)算數(shù)據(jù)提 供圖像顯示，用于檢驗(yàn)和編輯。The Graphical Display圖像顯示 The operator interface described here interacts withthe operator exclusively through a graphical in terface incorporating live video of the robot workspace 本操作界面可以讓操作者僅通過圖形界面與機(jī)器人工作空間真 實(shí)視頻畫面的結(jié)合進(jìn)行人

24、機(jī)交互。 A graphic model of the robot, with a viewpoint matching the video camera, is overlaid t o the video.機(jī)器人的圖像模型以與相機(jī)匹配的視角疊加到真實(shí)畫面上。Figure 4: The overlay graphical simulation.Figure 4 shows the graphical simulation of the RRC arm as a transparent image superimposed to the live video of the remote sit

25、e圖4所視是RRC手 臂以透明圖像疊加 到遙控地點(diǎn)的真實(shí) 畫面的圖形仿真Self-Collision Detection自身碰撞檢測(cè) Self-collision is defined as the situation when a manipulator link is too close to one of its other links. 自身碰撞被定義為一條操作桿與其他操作桿過于接近 的情況。 Approach: to test for line-segment t o line-segment distances, taking into account the radii of t

26、he links. 方法：檢測(cè)線段間的距離并考慮連桿的半徑Visualization of SingularitySpace奇異空間的可視化 The approach used here t o generate the joint angles and resolve the kinematic redundancy, given the set of end effector positions generated by the visual tracker, is the Configuration Control method這里采用生成關(guān)節(jié)角和解決運(yùn)動(dòng)學(xué)冗余的方法，由視覺 追蹤生成末

27、端執(zhí)行器位置。這就是配置控制方法。 Although configuration control prevents dangerous motions, it also alters the robot trajectory, possibly confusing the operator and creating problems in tight environments 盡管配置控制避免了危險(xiǎn)動(dòng)作，但他改變了機(jī)器人運(yùn)動(dòng)軌跡，可能會(huì)干擾操作者并在狹窄環(huán)境下產(chǎn)生問題。 To solve this problem we generate graphical warnings t o local

28、ize nearby singularities by using the distance from a singularity expressed as det(JJT)為了解決這個(gè)問題，我們生成了圖像警報(bào)，通過det()得到的 到奇異區(qū)域的距離確定是否靠近奇異區(qū)。 we visualize the singularity volume in the vicinity of the manipulator wrist based on the value of the determinant. 我們根據(jù)行列式值使執(zhí)行器附近的奇異區(qū)域可視化。結(jié)論和展望1 The innovative features of this system include a visual tracker to JH teleoperate a remote manipulator by simply moving the arms, and ar intelligent advisor, to analyze the commanded motions for kinematic correctness, enabling the operator to avoid self collision and singularities.該系統(tǒng)的創(chuàng)新點(diǎn)包括