智能轮椅导航与多传感器信息融合研究

【作者】 黄素平；

【导师】 何清华；

【作者基本信息】 中南大学 ， 机械电子工程， 2003， 硕士

【摘要】 智能轮椅是智能机器人技术在服务机器人领域的一个应用，目的是帮助老年人和残障人士提高行动自由度，让他们能重新融入社会。伴随着仿生学、微电子技术、模式识别等高科技的发展，轮椅功能发生了根本性的变革，具有记忆地图、避障、自动行走、与用户交互及为用户特殊需要服务等功能。本文综合分析了智能轮椅国内外研究现状，在此基础上对智能轮椅的导航及多传感器信息融合进行了研究。 本文首先提出了基于模块化设计思想的智能轮椅硬件系统和控制体系方案。对智能轮椅运动学特性进行了分析，建立了坐标变换数学模型。设计了基于模糊神经网络的多传感器信息融合过程。对BP网络的动量梯度下降算法和L-M优化算法进行了MATLAB仿真比较，提出了一种简单、高效的分区算法用于对传感器数据进行初步融合，并将障碍物距离信息模糊化，然后采用BP神经网络对障碍物环境分类，成功地利用L-M优化算法进行了模式识别。 在完成多传感器信息融合的基础上对智能轮椅导航进行了研究，实现了环境建模和路径规划：采用链接图法进行了全局环境信息建模，将全局路径规划转换为求带权无向图A中两点最短距离的问题，编制了基于C语言的Dijkstra算法程序实现全局路径规划，并提出了一种简单的几何方法优化全局路径：依据多传感器信息融合结果进行局部环境信息建模，实现了对环境的识别，设计了轮椅导航中跟踪全局路径、目标制导和避障行为，并给出了各种行为的算法。 开发了基于VC6.0的轮椅导航运动动态仿真平台，给出了多种情况下的仿真结果，仿真结果表明本文提出的导航方法是切实可行的，与其它已有方法相比具有在相近的障碍物之间可以发现路径和在障碍物前并没有出现振荡的优点。 以上的研究为智能轮椅的导航和多传感器信息融合提供了新的方法和思路，具有较好的规划能力、实时性和实用性。同时该研究方法也适用于移动机器人。更多还原

【Abstract】 Intelligent wheelchair is an application of intelligent robot technology . Wheelchairs as mobility aids to handicapped persons and the elderly help that they can be in harmony with society. With the development of bionics, micro-electronics and pattern distinction, the skill of wheelchair changes profoundly, having the abilities of remembering map,avoiding obstacles,autonomous mobile ,communicat-ing with users and servicing users. In this paper the navigation of intelligent wheelchair and multi-sensors fusion are researched.The hardware and soft structure of wheelchair are designed and the kinematics character is analyzed. A multi-sensors fusion system based fuzzy logic neural network is brought forward and the momentu grads descend arithmetic and L-M advanced arithmetic are compared. Furthmore,a subarea arithmetic for fusing primarily is put forward and the obstacle environment is sorted by BP network.Based the result of multi-sensors fusion the sort of environment and path planning are achieved. MAKLINK Graph is used for modeling whole environment and with Dijkstra arithmetic and a geometry method the mininum value is worked out . The behaviors among navigation are designed.A simulation system is designed through VC6.0. Simulation results in kinds of environment are given. The results indicate that the navigationmethod is good and feasible. Compared with other arithmetic the strongpoints of this method are it can find out safe path between near obstacles without surge.In the support of above achievement,a flexible and good method of navigation and multi-sensors fusion is found.The result in this paper can be used in mobile robot and high performance can also be gained.更多还原