摘要
轮腿式仿生机器人同时具有轮式机器人的高机动性和仿生足式机器人的强地形适应性，在灾后救援、野外勘探和太空探索等领域具有广泛应用，是机器人领域目前最前沿的研究之一。因此本文以实现轮腿式仿生机器人的自主性和智能化为研究目的，从以下几个方面展开了定位导航的研究。
动态场景下轮腿式仿生机器人定位算法研究。以提高高速动态场景下机器人的定位精度为目标，提出基于特征点速度的动态目标检测算法。建立并分析了前后图像间的旋转补偿模型和特征点像素速度模型，计算特征点速度后，通过运动滤波和投影滤波策略检测出动态目标上的特征点。
轮腿式仿生机器人定位算法初始化过程的优化。针对崎岖路面上轮腿式仿生机器人产生连续震动导致定位算法初始化失败问题，提出了基于视觉惯导信息融合的图像去运动模糊算法。研究了图像运动模糊产生的原因以及去运动模糊的原理，融合IMU测量得到的信息，推导了图像的运动模糊函数，复原了初始化时产生运动模糊的图像，减小了定位算法初始化失败的概率，优化了初始化过程。
let me in轮腿式仿生机器人导航算法研究。基于轮腿式仿生机器人腿部高度可变，提出具有三维可通过性的局部路径规划算法，并研究了常用的全局路径规划算法。通过仿真软件分别对全局路径规划算法与局部路径规划算法进行了测试分析。最后将全局路径规划算法与局部路径规划算法融合，形成轮腿式仿生机器人的导航算法。
定位导航算法验证实验。为了验证提出的动态目标检测算法、图像去运动模糊算法和基于腿部变高度的局部路径规划算法的有效性并测试各算法性能，搭建了轮腿式仿生机器人的定位导航系统，完成了实验平台的机械结构和硬件系统设计。在定位导航系统实验平台上分别进行运动滤波和投影滤波离线实验、动态目标检测实时实验、图像去运动模糊离线实验、局部路径规划避障实验以及全局路径规划导航实验。实验结果表明动态目标检测算法能够提高机器人在高速动态环境下的定位精度，去运动模糊算法能够提高定位导航算法初始化成功率，路径规划算法实现了基本导航任务。
关键词：轮腿式仿生机器人；定位导航；动态目标检测；图像复原
Abstract
The wheel-legged bionic robot has both the high mobility of the wheeled robot and the strong topographic adaptability of the bionic legged robot. It can be widely ud in post-disaster rescue, field exploration, space exploration and other fields. At prent, it is one of the most cutting-edge rearch in the field of robotics. In order to realize the autonomy and intelligence of the wheel-legged bionic robot, this paper studies the localization and navigation from the following aspects.
Rearch on localization algorithm of the wheel-legged bionic robot in the highly dynamic scene. A dynamic object detection algorithm bad on visual-inertial information fusion is propod to improve
the robot localization accuracy in the highly dynamic scene. The rotation compensation model and the feature point velocity model between the front and rear frames are established and analyzed. After calculating the feature point velocity, the feature points on the dynamic object are detected by motion filter and projection filter strategies.
Optimization of initialization process for the wheel-legged bionic robot localization algorithm. Aiming at the problem that the continuous vibration of the wheel-legged bionic robot on rough road leads to the failure of the initialization of the localization algorithm, an image deblurring algorithm bad on visual-inertial information fusion is propod. This paper studies the caus of image motion blur and the principle of motion blur removal and deduces the motion blur function of image with the measured information of the IMU. The motion blur image generated during the initialization is restored, which reduces the probability of localization algorithm initialization failure and optimizes the initialization process.
Rearch on navigation algorithm of wheel-legged bionic robot. The local path planning algorithm with the passability in three dimensions is propod bad on the variable leg height of the wheel-legged bionic robot. Meanwhile the commonly ud global path planning algorithm is studied. The global path planning algorithm and the local path planning algorithm are tested and analyzed respect
ively by simulation software. Finally, the global path planning algorithm and the local path planning algorithm are combined to form the navigation algorithm of the wheel-legged bionic robot.
Verification experiment of localization and navigation algorithm. In order to verify the effectiveness and performance of the propod dynamic object detection algorithm, image deblurring algorithm and local path planning algorithm bad on the variable leg height, the localization and navigation system of the wheel-legged bionic robot is built, and the mechanical structure and hardware system design of the experimental platform
are completed. Off-line experiments of motion filtering and projection filtering, real-time experiment of dynamic object detection, off-line experiments of image deblurring algorithm, obstacle avoidance experiment of local path planning and navigation experiment of global path planning are carried out respectively on the experimental platform of localization and navigation system. Experimental results show that the dynamic object detection algorithm can improve the accuracy of the robot localization in the highly dynamic environment, the deblurring algorithm can improve the initialization success rate of localization and navigation algorithm, and the path planning algorithm can achieve the basic navigation task.
sx是什么
Key words:wheel-legged bionic robot, localization and navigation, dynamic object detection, image restoration
jordana
目录
摘要 ............................................................................................................................... I I 目录 ............................................................................................................................ IV 第1章绪论 (1)
1.1 课题来源 (1)
1.2 课题背景及研究意义 (1)
1.3 国内外的研究现状及分析 (2)
1.3.1 视觉导航的仿生机器人的研究现状 (2)
1.3.2 移动机器人视觉SLAM的研究现状 (4)
1.3.3 移动机器人导航方式研究现状 (8)
1.4 主要研究内容 (9)
第2章双目相机与惯性测量单元联合标定 (11)
2.1 引言 (11)
2.2 双目相机标定 (11)
2.2.1 双目相机标定原理 (11)
2.2.2 双目相机标定实验 (15)奥斯卡金曲全集
2.3 惯性测量单元标定 (18)
网课平台哪个好2.3.1 惯性测量单元标定原理 (18)
2.3.2 惯性测量单元标定实验 (20)
2.4 双目相机与IMU联合标定 (21)
2.4.1 双目相机与IMU联合标定原理 (21)
2.4.2 双目相机与IMU联合标定实验 (22)
2.5 本章小结 (22)
第3章动态场景下轮腿式仿生机器人定位算法研究 (23)
3.1 引言 (23)
3.2 基于特征点速度的动态目标检测算法 (23)
3.2.1 动态目标检测算法框架 (23)
3.2.2 旋转补偿模型的建立与分析 (25)
3.2.3 特征点速度模型的建立与分析 (27)
3.2.4 双重滤波策略 (30)
heson3.3 基于视觉惯导信息融合的图像去运动模糊算法 (31)
3.3.1 图像去运动模糊原理 (31)
3.3.2 运动模糊核函数推导 (32)
3.4 本章小结 (34)越狱人物
第4章基于轮腿式仿生机器人的导航算法研究 (35)
4.1 引言 (35)
4.2 轮腿式仿生机器人导航算法框架 (35)
4.3 A*全局路径规划算法的研究 (35)
4.3.1 A*算法原理 (35)
4.3.2 全局路径规划仿真分析 (37)
4.4 具有三维可通过性的腿部变高度局部路径规划算法 (38)
4.4.1 腿部变高度局部路径规划算法原理 (38)
4.4.2 局部路径规划仿真分析 (42)
4.5 本章小结 (43)
第5章轮腿式仿生机器人定位导航实验 (44)
5.1 引言 (44)
5.2 轮腿式仿生机器人定位导航系统设计 (44)
family tree
5.2.1 机械结构设计 (44)
5.2.2 硬件系统设计 (45)
5.3 动态场景下轮腿式仿生机器人定位实验 (46)
5.3.1 运动滤波离线实验 (46)
5.3.2 投影滤波离线实验 (48)
小学英语教学案例
5.3.3 动态目标检测实时实验 (49)
5.3.4 图像去运动模糊实验 (53)
illustration5.4 轮腿式仿生机器人导航实验 (54)
5.4.1 局部路径规划避障实验 (54)
5.4.2 全局路径规划导航实验 (55)
5.5 本章小结 (56)
结论 (57)
