《外文翻译--移动机器人基于LFR激光探测器和IR的MFVFA方法.doc》由会员分享，可在线阅读，更多相关《外文翻译--移动机器人基于LFR激光探测器和IR的MFVFA方法.doc（15页珍藏版）》请在三一文库上搜索。
1、毕业设计(论文)外文资料翻译学院 (系)： 机械工程学院 专 业： 机械工程及自动化 姓 名： 学 号： (用外文写)外文出处： Mobile Robot Navigation Using Modified Flexible Vector Field Approach With Laser Range Finder and IR sensor 附 件： 1.外文资料翻译译文；2.外文原文。 指导教师评语：译文的意思基本正确，语句较通顺。专业性术语的翻译也较为得当。译文的数量已超过学校规定的要求。这说明该生具有较强的科技文献的阅读理解与翻译能力。 签名： 年 月 日注：请将该封面与附件装订成册。
2、附件1：外文资料翻译译文移动机器人基于LFR激光探测器和IR的MFVFA方法摘要： 在公共空间，移动的机器人可以用作导游者。指导一个人到达目标位置，移动的机器人的路径应该安全，可以避免障碍物并生成良好的导航的路径。一般来说，激光测距仪是用来检测在移动机器人周围的地图。我们建议移动机器人的导航方法用我们的，我们的方法可以在检测紧急情况下使用，它是我们开发的一种移动机器人的导航改性的柔性的矢量场与激光测距仪和红外传感器的方法，因为它高于激光测距仪响应的频率，通过实验结果表明了我们提出的控制方案和避障方法应用在公共场所里移动机器人的控制是非常有效的。关键词： 移动机器人导航 改性的柔性的矢量场方法
3、激光测距仪 红外传感器。1. 引言一个移动机器人可以在公共场所当向导，比如市场，邮局，图书馆等等，最重要的功能是弄够找到路径到达目标和导航的目标位置。无论怎样，在公共场所，移动机器人应该能够在避免障碍和到达目标位置同时进行。像一把椅子，一个架子和一个人等等这样的障碍。为达到导航目标位置的目的，许多研究者给了势场法的地址12。一个向量场柱状图3 5和动态窗口的方法6。由于市场是由许多狭窄的通道和许多障碍组成。以下是把势场法应用到市场里的机器人身上遇到的一些困难： 1)在近空间的障碍中很难找到通口。当机器人在狭窄的通道中移动会发生摆动运动。2)还有矢量场柱状图对环境地图的变化敏感，但是却不能找到到
4、达目标的路径，因为我们从移动机器人的完成中仅仅能得到角度的信息。动态窗口中使用了以目标,离障碍距离和速度为标题作为移动机器人的参数在动态窗口可以通过优化过程找到最佳速度。但是这不是唯一的为避免障碍而获得到达目标最短路径的方法。为了使机器人在公共场所中安全和稳定运行，一个新的导航方法是非常必要的，这种方法对于导航移动机器人环境的变化和运动最短路径的能力是敏感的。因此，我们提出一个新的移动机器人导航的方法，即移动机器人的导航改性的柔性的矢量场与激光测距仪和红外传感器的方法。在我们提出的方法中，由于路径信息的获得来自于障碍（作为圆）和移动机器人（作为一点）的几何关系，可以减少处理载荷。由于我们在移动
5、机器人中开发了具有差别驱动结构，当控制移动机器人趋向产生路径的时候，首先应该在运动学条件下考虑移动机器人稳定的速度。通常，如果把路径规划和路径跟踪分开，就会存在各种追踪控制方法，比如滑动模式，线性化，反演，神经网络，神经模糊系统。无论怎样，使用在传统的控制方法中，当追踪突然发生错误时，产生的这个基本速度命令是以极端大的估值和因遭受速度暴涨开始的。 在这篇论文中我们提出一个新的速度变化图的方法来保证移动机器人稳定的运动。如果我们第一时间假想移动机器人开始的位置和目标位置 ，在移动机器人的开始位置，目标位置和现在位置之间使用欧氏距离，我们就能生产参考的速度变化图。 2. 移动机器人平台和路径规划2
6、.1 移动机器人平台图1是我们开发的移动机器人平台。GIMAR, 移动机器人差别驱动结构有非完整约束。它有若干个传感器来检测移动机器人周围的状况。在这次研究中，我们仅仅使用两个传感器，一个是（激光测距仪）检测扫描地图的数据和另外一个是（红外传感器）紧急停止和避免障碍。这两个传感器各有优缺点。激光测距仪生产详细地扫描地图数据，但是它比红外传感器运行的慢。另外一方面，红外传感器比激光测距仪运行迅速除了它生成一点数据。因此，我们打算提出的是为稳定驱动控制和避障这两个传感器的数据相结合的方法。2.2 路径规划为了到达目标位置，在它开始移动之前我们就应该知道它的路径了。如果地图数据时提前给出，我们就能生
7、成安全路径到达目标的立场，否则我们就不能得到完整的路径。在文中，我们假设如下：1.地图数据时没有提前给出。2.仅仅给出了移动机器人的初始位置和目标位置。3.移动机器人没有滑运动，我们可以从里程表的信息中知道移动机器人的的位置和移动机器人的姿势由以上三个假设，我们提出了路径规划策略在图2中表明。图2 危险的区域指示了移动机器人没有障碍碰撞的一个区域。我们就可以直观地知道红线边沿的路径是最短的安全的路径。我们事先没有地图数据，我们就建议找出下一个预期的点，这一点来自危险区域的圆和移动机器人现在位置的几何关系。虽然障碍到处都是，但是从我们提出的改性的柔性的矢量场的方法中我们就可以决定下一个预期的点，
8、在扫描期间里路径规划过程不断的被废除。3. 移动机器人控制非完整移动机器人可以有两个坐标系统，一个是由XG,YG ,SG构成的世界坐标系和另一个是由XL,YL,6L构成的本地坐标系.在图3中，C是机器人中心点，D是做齿轮和右齿轮的距离，B是齿轮中心和主销后倾的距离。V和W意思是移动机器人的线速度和角速度。自由移动的移动机器人是作为完整的移动机器人有三个自由度（X,Y,Z和零点）。但是，由于运动学上的限制，非完整移动机器人的自由度减少到两个。在没有滑动的条件下，一个非完整移动机器人的运动限制由以下公式给出从运动控制角度看，我们开发的移动机器人有vC线速度和wc角速度的两个自由度，把这两个齿轮的直
9、径，半径，角速度描绘成两轮的速度（wL, wR.）。可以用以下两轮的角的速度关系来叙述线速度和角速度。为了控制移动机器人，两个齿轮的速度可以分成两部分：一个是确定移动机器人的线速度和另一个是追踪移动机器人的姿态。vl = vc - wc vR = vc + wc vc 是线速度控制部分和wc是角速度控制部分。线速度控制的目标是根据移动机器人的位置和目标位置的距离来控制移动机器人的速度。在接下来的分段中，我们提出用欧氏距离来生成线速度的方法。移动机器人通过控制wc能够避免障碍。在接下来的章节中，我们将解释利用MFVFA对姿态误差反馈的方法4. 仿真和实验结果由于提出避障运算法则要用在狭窄的市场空
10、间中为目的，就在通道中设置像架子之类的物体作为测试环境。我们组成了市场模型的两部分。这个测试环境的尺寸是由6米*6米。我们假设没有人在测试环境中。在实验结果中，点线指示了要求的位置和红线指示了激光扫描数据。图表中也展示了多障碍情况。图11是我们看到这个实验结果。如果在测试环境中有多障碍，这时提出的运算法则能够提供安全通道，保证移动机器人无碰撞地通过通道。之后，那个运算法则生成到达要求位置的最短轨迹图.10图.10稳定的曲率跟踪运算法则仿真结果（a）右拐弯仿真的结果 （b）左拐弯仿真的结果图.10展示稳定的曲率跟踪运算法则仿真结果。这个结果表明了当机器人进入了危险区域是机器人是怎样逃离危险区域的
11、。无论机器人的位置在哪里，是在危险区域还是在碰撞区域，在危险区域圆周围都可以控制机器人。图.10表明了，机器人利用柔性的矢量场根据机器人的位置直接脱离危险区域。因此，移动机器人移动到安全区域和接受稳定的最短的路径到达目标位置。 图.11图.11： 机器人和多障碍之间避免碰撞的结果（a）-在运用运算法则前机器人和多障碍之间避免碰撞的结果（b）-在运用运算法则后机器人和多障碍之间避免碰撞的结果图.12 展示了机器人轨迹。图中的数据来自图11实验的结果。图12展示了机器人安全地避免障碍的结果。我们设定F200,10Q), F(150,120), P(200,100), P(16G,180)作为开始位
12、置，设定P(240,450), P(I30,500), (230,500), (170,600)为要求的位置。我们把机器人的速度设成70cm/s与人类步行速度一样.我们成功地通过在开始位置和要求的位置之间随意地设定障碍的狭窄的通道。图.12 展示了机器人避开障碍物。图.12：在运用避障运算法则后机器人的轨迹(a)机器人的轨迹从（200，100）到(240，450Cm),(b)机器人的轨迹从（180，160）到(160，520Cm),(c)机器人的轨迹从（200，100）到(230，500Cm),(d)机器人的轨迹从（160，180）到(170，600Cm),5. 结论在这篇论文中，我们讨论了导
13、航方法关于在改性的柔性的矢量场中使用激光测距仪和红外传感器的方法。该控制器分为线速度控制和角速度控制部分。利用欧氏距离并考虑了移动机器人在稳定运动时而生成线速度剖面图 。角速度部分，我们利用了虚拟圆起源于角点和切向直线。无论怎样，因为移动机器人存在碰撞区域，所以我们通过利用提出的稳定曲率运算法则来控制移动机器人。附件2：外文原文Mobile Robot Navigation using Modified Flexible Vector Field Approach with Laser Range Finder and IR sensorJinpyo Hong ,Youjun Choi and
14、 Kyihwan ParkAbstract: In the public space, a mobile robot is adopted as a guider. For guiding a person to the goal position, the mobile robot should make the safe path ,avoid the obstacles and navigate the generated path well. In general, Laser Range Finder is used for the detection of the map arou
15、nd the mobile robot. We propose mobile robot navigation method using our developed a Modified Flexible Vector Field Approach with Laser Range Finder and IR sensor which is used for detecting the emergency status because it has higher response frequency than that of LRF. We will verify that our propo
16、sed control scheme and obstacle avoidance algorithm are useful enough to apply to the mobile robot control in the public space by showing experimental results. Keywords: Mobile robot navigation, MFVFA(Modified Flexible Vector Field Approach, LRF(Laser Range Finder), IR sensor)1. INTRODUCTIONWhen a m
17、obile robot is operated as a guider in the public spaces such as a market, a post office, a museumand so on, the most important functionality is the ability to find the path to reach a goal and navigate the goal position. However, In the public space, since there are many obstacles like a shelf, a c
18、hair and a person, the mobile robot should avoid the obstacles and reach the goal position simultaneously. For the purpose of navigate the target position, many researchers have addressed a potential field method12, a vector field histogram3 5 and a dynamic window approach6. Since the market is comp
19、osed of many narrow passageways and a lot obstacles, there are some difficulty to apply the potential field into our robot as a market application as follows: 1) It is hard to find the passage in the close spaced obstacles. 2) There is an oscillation motion when the robot moves in the narrow passage
20、ways. Also the vector field histogram is sensitive to the change of the environmental map but it can not find the shortest path to reach the goal because we only get the angle information for the mobile robot to go through. The dynamic window uses the heading to goal, distance to obstacles and the v
21、elocity of the mobile robot as parameters. Since the dynamic window has an optimisation process for finding the best velocity, this solution is not unique to the shortest path for avoiding the obstacle. For the safe and stable operation of the mobile robot in the public space, a new navigation metho
22、d is essentially needed which is sensitive to the change of the environment and capable of moving in the shortest path as a guider. Therefore, we proposed a new mobile robot navigation method using modified flexible vector field approach with Laser Range Finder(LRF) and Infra Red(IR) sensor. In our
23、proposed method, since the path informationis obtained from the geometric relation of the obstacle as a circle and the mobile robot as a point, the processing load can be decreased. Since our developed mobile robot has a differential drive structure, the stable velocity of the mobile robot should be
24、 firstly considered in this kinematic condition when the mobile robot is controlled toward the generated path. Usually, if the path planner and the path tracker are divided, the various tracking control method are existed like as sliding mode, linearization, backstepping, neural networks, neuro-fuzz
25、y systems. However, the generated root velocity command using those conventional control approaches start with a very large value, and suffers from velocity jumps when sudden tracking errors occur. Therefore, we propose a new velocity profiling approach guaranteeing the stable movement of the mobile
26、 robot in this paper. If we assumed that the initial position of the mobile robot and the goal position are given at the first time, we can generate the reference velocity profile using euclidean distances among the start position , the goal position and the current mobile robot position.2. MOBILE R
27、OBOT PLATFORM AND PATH PLANNING2.1 Mobile robot platform Figure 1 is our developed mobile robot platform, GIMaR. The mobile robot has a differential drive structure that has non-holonomic constraint. It has several sensors for detecting the status around the mobile robot. In this research, we only u
28、sed two sensors that one is LRF for detecting the scan map data and another is IR sensor for emergency stop and obstacle avoidance. These two sensors have advantages and drawbacks respectively. LRF generates the scan map data in detail but it is operated slowly compared with IR sensor. On the other
29、hand, IR sensor operates rapidly compared with LRF but it generates one point data. Therefore, we intend to propose the method combining two sensors data for stable driving control and obstacle avoidance.2.2 Path planningIn order lo reach the goal position, we should know he path before starting lo
30、move. If the map data is given in advance, we can generate the safe path to arrive at the goal position but otherwise, we can not do the entire path. In this paper, we assume as follows.1 .Map data is not given in advanced.2. Only the initial position of the mobile robot and the goal position are gi
31、ven.3. Mobile robot has no slip motion and we can know the position and posture of the mobile robot from the odometry information.Under three assumptions, our proposed path planning strategy is shown in Fig. 2. In Fig. 2. the dangerous region indicates an area that the mobile robot has no collision
32、with obstacles. We can know intuitively (hat red line path is the shortest path in safely. Since we dont have map data in advance, we suggest that we find out the next desired point from the geometric relation of dangerous region circle and the current mobile robot position. Although the obstacle ex
33、ists anywhere, we can determine the next desired point from the our proposed approach, modified flexible vector field approach(MFVFA) because the path planning process is repealed during scan period continuously.3. MOBILE ROBOT CONTROLA nonholonomic mobile robot can be represented by two coordinate
34、systems which are the world coordinate system XG,YG ,SG and the local coordinate system Xl,Yl,6L,. In figure 3, C is the robot center point, d is the distance between the left wheel and the right wheel and b is the distance between the center of the wheel and the caster. V and w mean the linear velo
35、city and the angular velocity of the mobile robot.A freely movable mobile robot that is referred as holo-nomic mobile robot has three degrees of freedom(d.o.f.) X, Y, and 0, However, because of the kinematical constraint, the degrees of the freedom for a nonholonomic mobile robot reduces to two. On
36、the conditions of non-slipping, the kinematic constraint of a nonhomolonomic mobile robot is given asFrom the motion control perspective, our developed mobile robot has 2 d.o.f., vC, wc , where vc is the linear velocity and wc is the angular velocity of the mobile robot.The velocities of the two whe
37、els are represented as the diameter of the wheel, r, and the angular velocity wL, wR. .The linear velocity and angular velocity can be described as the relation of e both wheels angular velocities as follows.In order to control the mobile robot, the velocities of the two wheels can be divided us two
38、 parts: one is the part determining the linear velocity of the mobile robot and another is the part tracking the posture of the mobile robot.vl = vc - wc vR = vc + wc Where, vc is the linear velocity control part and wc is the angular velocity control part. The objective of the linear velocity contr
39、ol is the velocity control of the mobile robot according to the distance between the robot position and the goal position. In the next subsection, we propose the linear velocity generation method using euclidean distance. The mobile robot can avoid the obstacle by controlling the wc. We will explain
40、 the posture error feedback method using MFVFA in the next section.4. SIMULATION AND EXPERIMENTAL RESULTSSince we have suggested the obstacle avoidance algorithm for the purpose of using in the narrow space at the market, shelf-like objects are set in the passageway for a test environment. We made u
41、p two sections of the market model as a test environment. The size of the test environment is 6 meter by 6 meter. We assume that there are no people in the test environment. In the experimental results, dotted line indicates the desired position, and the red line indicates laser scanning data.The re
42、sult of multi-obstacle case is also shown in Fig. 11. As we see this experimental result in Fig. 11, if there are multi-obstacles in the test environments, the proposed algorithm provides the safe passageway that the robot can go through without collision. After that, the algorithm generates the sho
43、rtest trajectory to approach the desired position. Fig. 10Fig. 10 Simulation Result of Stable Curvature Tracking Algorithm - a) the Right Turning Simulation Result, b) the Left Turning Simulation ResultFig. 10 show simulation results of the stable curvature tracking algorithm. These results show how
44、 the robot escapes dangerous region when the robot moves into the dangerous region. Wherever the robot is positioned between the dangerous region and collision region, the robot can be controlled toward the circumference of the dangerous region circle. As shown in Fig. 10, the robot makes flexible v
45、ector fields with the direction to out-of-dangerous region according to the location of the robot. Therefore, the mobile robot moves to the safe region and follows the stable and shortest path for arriving at the goal position. Fig. 11Fig 11 : Result of the Collision Avoidance between Robot and Mult
46、i-Obstacles - (a) before applying the collision avoidance algorithm between robot and multi-obstacles (b) after applying the collision avoidance algorithm between robot and multi-obstaclesThe Trajectories of the robot are shown in Fig. 12 obtained from the experimental results which are shown in Fig
47、. 11. Fig. 12 shows the result that the robot avoids Obstacles safely. We set F200,10Q), F(150,120), P(200,100), P(16G,180) as starting positions and P(240,450), P(I30,500), (230,500), (170,600) as desired positions. We move the robot with speed or 70cm/s which is same as the walking speed of the human. We made narrow passageway by adding arbitrary obstacles between starting position and desired position. Since there is an obstacle, the robot avoids obstacles, as shown in Fig. 12. Fig. 12Fig. 12: Trajectorie