【作者】 陈少斌；

【导师】 蒋静坪；

【作者基本信息】 浙江大学 ， 控制理论与控制工程， 2008， 博士

【摘要】 移动机器人是近年来发展起来的一门综合学科,集中了机械、电子、计算机、自动控制以及人工智能等多学科的最新研究成果,代表了机电一体化的最高成就。近年来,移动机器人已经成为自动化、计算机和人工智能等领域的一个研究热点。和传统的工业机器人相比,具有自主感知、决策和执行功能的移动机器人具有广阔的应用前景,在国防、工农业生产、抢险等领域中都具备着人类所无法比拟的巨大优势。本文以自主移动机器人为研究对象,针对它的路径规划及轨迹跟踪问题,做了以下研究工作:1.简述了机器人的发展历史和移动机器人的特点,较为系统、全面地综述了移动机器人的体系结构、多传感器信息融合、定位技术、路径规划以及轨迹跟踪控制等技术。2.提出一种基于异质传感器信息融合的机器人自定位方法。首先,建立机器人运动模型和CCD摄像机观测模型,利用扩展卡尔曼滤波器进行状态观测和抑制模型的随机噪声,并实现自主移动机器人的自定位。然后,建立超声波传感器的观测模型,借助于BP神经网络,将CCD摄像机获取的自定位信息与超声波传感器获取的自定位信息进行融合,实现两种传感器的优缺点互补。仿真结果表明,通过这种异质传感器的信息融合后,自主移动机器人的自定位精度得到明显的提高。3.针对特定的静态环境,提出了一种改进的十进制可变长度编码机制,并设计出了对应于该编码机制的遗传算子,将遗传算法应用于静态环境下移动机器人的路径规划,仿真实验验证了算法的有效性。简要叙述了基本蚁群算法的原理,研究了自适应蚁群算法在移动机器人路径规划中的应用。4.结合神经网络,将粒子群优化算法应用于动态环境下移动机器人的路径规划。用神经网络模型描述机器人工作空间的动态环境信息,并建立起机器人动态避障与网络输出间的关系,将路径的二维编码简化为一维编码,然后用粒子群优化算法获得最优无碰路径。介绍了距离信息传播算法的基本原理,以追捕运动目标为目的,在动态环境中,运用距离信息传播算法快速规划出了移动机器人的最优路径。5.研究了具有量测噪声、模型噪声和输入信号干扰的四轮移动机器人轨迹跟踪控制。建立了系统的运动学与动力学方程和轨迹生成方程,采用卡尔曼滤波器的状态估计,提出一种基于李亚普诺夫稳定性的最优状态反馈控制策略,设计了补偿输入信号干扰的自适应算法,给出仿真结果,验证了所提算法的有效性。更多还原

【Abstract】 Mobile robot system, as an integrative subject with updated research results in mechanical, electronic, computer, automatic control, and artificial intelligence, represents tremendous success of mechatronics. In recent years, mobile robot system has become an important research area in automation, computer, and artificial intelligence. Comparing with traditional industrial robot, the mobile robot with self-perception, decision making, and performance function has vaster application prospect. Mobile robot has huge advantages over human being in the fields of national defense, industrial and agriculture manufacture, and hazard. This dissertation studies the path planning and trajectory tracking of an autonomous mobile robot, it mainly contains:1. The development history of robot and the characteristic of mobile robot are briefly narrated. The technology of the architecture of mobile robot, multi-sensor information fusion, self-localization, path planning and trajectory tracking are summarized in the round.2. A self-localization method in the presence of measurement noise and model disturbance for autonomous mobile robot based-on heterogeneous sensor information fusion is proposed. First, the motion model of the autonomous mobile robot and observed model of CCD vidicon are established. The optimal state estimation is derived, random noises are overcome, and self-localization is realized by extended Kalman filter. Then, observed model of the ultrasonic sensor is established. The data from CCD vidicon and ultrasonic sensor are fused by BP neural network. The mutual aided of the tow kinds of sensors is realized. The simulation results show that the self-localization precision of the autonomous mobile robot is obviously improved by heterogeneous sensor information fusion. 3. Under the specific environments, a novel variable-length decimal encoding mechanism for the paths of the mobile robot is proposed, some genetic operators corresponding to the encoding mechanism are designed at the same time, and genetic algorithm is utilized to program an optimal path of mobile robot in the static environment, simulation results demonstrate the effectiveness of the algorithm. Principle of the basic ant colony algorithm is briefly described, and path planning for mobile robot based on the self-adaptive ant colony algorithm is studied.4. Combining the neural network, particle swarm optimization algorithm is utilized to program an optimal path of mobile robot in the dynamic environment. The dynamic environmental information in the workspace for a robot is described by a neural network model, using this model, the relationship between dynamic obstacle avoidance and the output of the model is established, then the two-dimensional coding for the planned path is simplify to one-dimensional one, and the particle swarm optimization is introduced to get an optimized collision-free path. Then, the introduction of the basic principle of distance-propagation algorithm is given, aiming at pursuing the locomotive target, by utilizing the distance-propagation algorithm, we can program the optimal path of mobile robot in the dynamic environment quickly.5. The optimal state feedback control for trajectory tracking control of a four-wheel mobile robot is studied in the presence of measure noise, model noise and input signal noise as well. At first, the kinematic and dynamic model and trajectory generation of the omnidirectional vehicle have been established. Then an optimal state feedback controller based on Lyapunov stability using the Kalman filter state estimation technique is derived. This is followed by an adaptive control algorithm to compensate for the effects of input signal noise. Simulated results are presented in the paper to highlight the effectiveness of the proposed control algorithm.更多还原